002_Data_Science_Challenge_2020.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.028529,
     "end_time": "2020-08-30T09:33:38.177402",
     "exception": false,
     "start_time": "2020-08-30T09:33:38.148873",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Data Science Challenge 2020\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.026448,
     "end_time": "2020-08-30T09:33:38.230878",
     "exception": false,
     "start_time": "2020-08-30T09:33:38.204430",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Inhaltsverzeichnis:\n",
    "* [Einleitung](#Einleitung)\n",
    "* [Der verwendete Datensatz](#Datensatz)\n",
    "* [Übersetzung von R nach Python](#VonRNachPython)\n",
    "    - [Definition der Trainings- bzw. Testdaten](#DefinitionTrainingsTestdaten)\n",
    "    - [GLM - Generalized Linear Models](#GLM)\n",
    "    - [CART - Classification and Regression Trees](#CART)\n",
    "    - [BAGGING - Bootstrap Aggregation](#Bagging)\n",
    "    - [RF - Random forests](#Randomforests)\n",
    "    - [XGB - Gradient Boosting](#GradientBoosting)\n",
    "    - [BART - Bayesian Additive Regressions Trees](#BART)\n",
    "    - [MARS - Multivariate Adaptive Regression Splines](#MARS)\n",
    "    - [Fazit](#Fazit)\n",
    "* [Erweiterung der Modellvielfalt um ein neuronales Netz](#ErweiterungNeuronalesNetz)\n",
    "* [Zusammenfassung der Ergebnisse](#Zusammenfassung)\n",
    "* [Links](#Links)"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.026231,
     "end_time": "2020-08-30T09:33:38.283809",
     "exception": false,
     "start_time": "2020-08-30T09:33:38.257578",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "<a id=\"Einleitung\"></a>\n",
    "# Einleitung\n",
    "Aufgrund des [Vortrags](https://www.actuview.com/data-science-challenge-2020_354ff90b8.html) von Friedrich Loser im Rahmen der e-Jahrestagung zum Thema 'Data Science Challenge 2020' sind wir auf einen öffentlich zugänglichen Datensatz aufmerksam geworden, der von Mervyn Kopinsky im Rahmen der Studie ['Predicting Group Long Term Disability Recovery and Mortality Rates Using Tree Models'](https://www.soa.org/globalassets/assets/files/research/projects/2017-gltd-recovery-mortality-tree.pdf) verwendet wurde. Dieser Datensatz wurde nun auch in einer weiteren Studie ['Machine-Learning Methods for Insurance Applications - A Survey'](https://www.soa.org/globalassets/assets/files/resources/research-report/2019/machine-learning-methods.pdf) von Alex Diana, Jim E. Griffin, Jaideep Oberoi und Ji Yao verwendet und durch weitere, aus den gegebenen Daten abgeleitete Größen erweitert. Diese Daten dienen auch als Grundlage für diese Arbeit. \n",
    "\n",
    "Die Studie 'Machine-Learning Methods for Insurance Applications' widmet sich der Fragestellung, wie man neuartige Entwicklungen in der Sammlung von Daten sowie deren Auswertung im Rahmen von 'Data Science' in der aktuariellen Welt nutzbar machen kann. Denn aufgrund wachsender Datenmengen ist es nötig Verfahren zu entwickeln, mit deren Hilfe man diese Datenmenge effizient auswerten kann, indem relevante Variablen in Vorhersagemodellen ausgewählt und falsche Zusammenhänge identifiziert und beseitigt werden. Schadensdaten sind in der Regel spärlich, flüchtig, verzerrt und zeitlich variabel. Dies macht sie zu hervorragenden Kandidaten für die Analyse mit maschinellen Lernansätzen, insbesondere mit variablen Auswahlmethoden. \n",
    "    \n",
    "Neben der Studie 'Machine-Learning Methods for Insurance Applications' selbst wurde auch der verwendete Datensatz inklusive des Jupyter-Notebook veröffentlich, so dass diese Arbeit erst möglich gemacht wurde. Die Ergebnisse der Studie basieren auf der Sprache 'R', sodass das erstes Ziel dieser Arbeit ist, die im Jupyter-Notebook beschriebenen Rechnungen in der Programmiersprache 'Python' wiederzugeben und die Ergebnisse mit denen der Studienautoren zu vergleichen. Wir gehen davon aus, dass keine signifikanten Unterschiede zu erkennen sein werden - oder anders formuliert, dass die Sprachen 'R' und'Python' gleichwertige Ergebnisse liefern. \n",
    "\n",
    "Neben diesen vorgestellten Methoden beschäftigen wir uns weiterhin mit einem Random Forest. Innerhalb des Vergleichs zwischen den beiden Programmiersprachen wird ebenfalls ein Modell dieser Art verwendet. Bei einem Random Forest handelt es sich um eine weit verbreitete Methode, die im Rahmen der Analyse von Daten und den daraus abgeleiteteten Vorhersagen benutzt wird. Wir wollen daher einige Schritte vorstellen, die man ergreifen kann, um eine bessere Vorhersage zu erhalten.\n",
    "\n",
    "Die Studie untersucht und vergleicht verschiedene Machine-Learning-Methoden miteinander. Dabei fällt auf, dass künstiche neuronale Netze nur sehr kurz behandelt werden. Die Autoren haben ein einfaches neuronales Netz untersucht, welches im Vergleich zu den anderen Methoden schlechte Ergebnisse liefert, stellen jedoch heraus, dass künstliche neuronale Netze in der Vergangenheit bereits sehr gute Ergebnisse erzielt haben. Unser zweites Ziel ist daher die Entwicklung eines künstlichen neuronalen Netzes, welches eine mindestens ähnliche Güte im Vergleich zu den anderen Machine-Learning-Methoden aufweist. Die Ergebnisse des neuronalen Netzes werden abschließend mit den Ergebnissen der Studie verglichen.\n",
    "\n",
    "Diese Arbeit gliedert sich demnach in die drei Ziele:\n",
    "1. Übersetzung der bereits untersuchten Machine-Learning Methoden von 'R' nach 'Python'.\n",
    "2. Untersuchung des Random-Forests auf Verbesserungsmöglichkeiten.\n",
    "3. Aufbau und Training eines künstlichen neuronalen Netzes. \n",
    "\n",
    "Zunächst laden wir alle Pakete, die wir für unsere Analysen benutzen:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:33:38.356475Z",
     "iopub.status.busy": "2020-08-30T09:33:38.355694Z",
     "iopub.status.idle": "2020-08-30T09:33:52.568237Z",
     "shell.execute_reply": "2020-08-30T09:33:52.567488Z"
    },
    "papermill": {
     "duration": 14.257893,
     "end_time": "2020-08-30T09:33:52.568364",
     "exception": false,
     "start_time": "2020-08-30T09:33:38.310471",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Collecting gbart\r\n",
      "  Downloading gbart-0.2.1-py3-none-any.whl (40 kB)\r\n",
      "     |████████████████████████████████| 40 kB 175 kB/s \r\n",
      "\u001b[?25hRequirement already satisfied: matplotlib in /opt/conda/lib/python3.7/site-packages (from gbart) (3.2.1)\r\n",
      "Requirement already satisfied: numpy in /opt/conda/lib/python3.7/site-packages (from gbart) (1.18.1)\r\n",
      "Requirement already satisfied: joblib in /opt/conda/lib/python3.7/site-packages (from gbart) (0.14.1)\r\n",
      "Requirement already satisfied: sklearn in /opt/conda/lib/python3.7/site-packages (from gbart) (0.0)\r\n",
      "Requirement already satisfied: statsmodels in /opt/conda/lib/python3.7/site-packages (from gbart) (0.11.1)\r\n",
      "Requirement already satisfied: seaborn in /opt/conda/lib/python3.7/site-packages (from gbart) (0.10.0)\r\n",
      "Requirement already satisfied: pandas in /opt/conda/lib/python3.7/site-packages (from gbart) (1.0.3)\r\n",
      "Requirement already satisfied: scipy in /opt/conda/lib/python3.7/site-packages (from gbart) (1.4.1)\r\n",
      "Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /opt/conda/lib/python3.7/site-packages (from matplotlib->gbart) (2.4.7)\r\n",
      "Requirement already satisfied: kiwisolver>=1.0.1 in /opt/conda/lib/python3.7/site-packages (from matplotlib->gbart) (1.2.0)\r\n",
      "Requirement already satisfied: python-dateutil>=2.1 in /opt/conda/lib/python3.7/site-packages (from matplotlib->gbart) (2.8.1)\r\n",
      "Requirement already satisfied: cycler>=0.10 in /opt/conda/lib/python3.7/site-packages (from matplotlib->gbart) (0.10.0)\r\n",
      "Requirement already satisfied: scikit-learn in /opt/conda/lib/python3.7/site-packages (from sklearn->gbart) (0.23.1)\r\n",
      "Requirement already satisfied: patsy>=0.5 in /opt/conda/lib/python3.7/site-packages (from statsmodels->gbart) (0.5.1)\r\n",
      "Requirement already satisfied: pytz>=2017.2 in /opt/conda/lib/python3.7/site-packages (from pandas->gbart) (2019.3)\r\n",
      "Requirement already satisfied: six>=1.5 in /opt/conda/lib/python3.7/site-packages (from python-dateutil>=2.1->matplotlib->gbart) (1.14.0)\r\n",
      "Requirement already satisfied: threadpoolctl>=2.0.0 in /opt/conda/lib/python3.7/site-packages (from scikit-learn->sklearn->gbart) (2.1.0)\r\n",
      "Installing collected packages: gbart\r\n",
      "Successfully installed gbart-0.2.1\r\n",
      "WARNING: You are using pip version 20.1.1; however, version 20.2.2 is available.\r\n",
      "You should consider upgrading via the '/opt/conda/bin/python3.7 -m pip install --upgrade pip' command.\r\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd \n",
    "seed_value= 1\n",
    "import os\n",
    "os.environ['PYTHONHASHSEED']=str(seed_value)\n",
    "import random\n",
    "random.seed(seed_value)\n",
    "import numpy as np\n",
    "np.random.seed(seed_value)\n",
    "import tensorflow as tf\n",
    "tf.random.set_seed(seed_value)\n",
    "from tensorflow import keras\n",
    "from keras.wrappers.scikit_learn import KerasClassifier\n",
    "import kerastuner as kt\n",
    "import IPython\n",
    "import matplotlib.pyplot as plt\n",
    "import math\n",
    "import statsmodels.formula.api as smf\n",
    "import statsmodels.api as sm\n",
    "import datetime\n",
    "import xgboost as xgb\n",
    "import graphviz\n",
    "import pyearth\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.model_selection import RandomizedSearchCV\n",
    "from sklearn.ensemble import RandomForestRegressor\n",
    "from sklearn.metrics import mean_squared_error\n",
    "from sklearn import tree\n",
    "from pprint import pprint\n",
    "from xgboost import plot_tree\n",
    "!pip install gbart\n",
    "import gbart\n",
    "from gbart.modified_bartpy.sklearnmodel import SklearnModel"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.027836,
     "end_time": "2020-08-30T09:33:52.624168",
     "exception": false,
     "start_time": "2020-08-30T09:33:52.596332",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "<a id=\"Datensatz\"></a>\n",
    "# Der verwendete Datensatz\n",
    "In diesem Abschnitt wollen wir uns mit den Variablen der beiden Modelle beschäftigen, um zu erläutern welche Art von Daten in die Modelle eingeflossen sind. Es ist wichtig zu verstehen, welche Daten vorliegen, um entsprechende Modelle fundiert zu definieren und zu analysieren.\n",
    "In beiden Fällen liegen bereits strukturierte Datensätze vor, so dass wir die Daten an dieser Stelle nicht zunächst in eine strukturierte Form überführen müssen. Wir können also direkt dazu übergehen den Datensatz spaltenweise zu betrachten und die in den jeweiligen Spalten enthaltenen Informationen zu beschreiben.\n",
    "\n",
    "Im Rahmen dieser Arbeit benutzen wir die gegebenen Daten aus der Studie, ohne diese auf ihren kausalen Zusammenhang zu untersuchen. Wir nehmen an, dass die Daten, die zur Bildung der Modelle aus dem nächsten Abschnitt verwendet werden, bereits hinreichend geprüft wurden. D.h. wir gehen davon aus, dass die benutzten Variablen signifikanten Einfluss auf die Ergebnisse des Modells haben, sowie alle nicht verwendeten Daten diesen Einfluss nicht aufweisen. \n",
    "\n",
    "Desweiteren haben wir die Qualität der Daten nicht geprüft. Denn wie in der Einleitung bereits beschrieben ist ein Ziel dieser Arbeit, die Modelle, die in der Sprache 'R' geschrieben wurden, gegen identische Modelle in der Sprache 'Python' zu testen. Ebenfalls ist für die Erweiterung durch ein künstliches neuronales Netz und den angestrebten Vergleich mit den existierenden Modellen entscheidend, dass die gleiche Datengrundlage verwendet wird.\n",
    "\n",
    "Dennoch möchten wir an dieser Stelle zeigen, wie die Ausprägungen der verwendeten Variablen beschrieben sind. Dazu verwenden wir gezielt zur Darstellung passend gewählte Visualisierungsformen. Wir betrachten dabei nur die Variablen, die für die Modelle des nächsten Abschnitts relevant sind.\n",
    "\n",
    "Wir lesen den Datensatz für die weitere Bearbeitung ein. Als Grundlage dieser Arbeit dient der frei verfügbare Datensatz 'recovery', der zur Studie ['Machine-Learning Methods for Insurance Applications - A Survey'](https://www.soa.org/globalassets/assets/files/resources/research-report/2019/machine-learning-methods.pdf) gehört und [hier](https://www.soa.org/resources/research-reports/2019/machine-learning-methods/) heruntergeladen werden kann. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "_cell_guid": "b1076dfc-b9ad-4769-8c92-a6c4dae69d19",
    "_uuid": "8f2839f25d086af736a60e9eeb907d3b93b6e0e5",
    "execution": {
     "iopub.execute_input": "2020-08-30T09:33:52.693679Z",
     "iopub.status.busy": "2020-08-30T09:33:52.688376Z",
     "iopub.status.idle": "2020-08-30T09:33:54.730573Z",
     "shell.execute_reply": "2020-08-30T09:33:54.729739Z"
    },
    "papermill": {
     "duration": 2.078554,
     "end_time": "2020-08-30T09:33:54.730704",
     "exception": false,
     "start_time": "2020-08-30T09:33:52.652150",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# Einlesen der Dateien\n",
    "recovery = pd.read_csv('../input/gltdtestdata/2017-gltd-tree-recovery-rate-input-data.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.027191,
     "end_time": "2020-08-30T09:33:54.785786",
     "exception": false,
     "start_time": "2020-08-30T09:33:54.758595",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Im folgenden wollen wir uns dem verwendeten Datensatz ein wenig nähern und diesen zunächst analysieren. Dabei betrachten wir die Größe des Datensatzes und deren einzelne Parameter inkl. der zugehörigen Ausprägungen."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:33:54.848598Z",
     "iopub.status.busy": "2020-08-30T09:33:54.847475Z",
     "iopub.status.idle": "2020-08-30T09:33:54.853688Z",
     "shell.execute_reply": "2020-08-30T09:33:54.854725Z"
    },
    "papermill": {
     "duration": 0.041569,
     "end_time": "2020-08-30T09:33:54.854928",
     "exception": false,
     "start_time": "2020-08-30T09:33:54.813359",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\"Anzahl der Datenätze in 'recovery' entspricht: 818941\"\n"
     ]
    }
   ],
   "source": [
    "# Anzahl der Datensätze\n",
    "pprint(\"Anzahl der Datenätze in 'recovery' entspricht: \" + str(len(recovery)))"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.027601,
     "end_time": "2020-08-30T09:33:54.910479",
     "exception": false,
     "start_time": "2020-08-30T09:33:54.882878",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Der Datensatz besitzt die folgendenen Spalten mit den nachfolgenden Ausprägungen. Zu Details zu der Erhebung, dem Preprocessing und der Bedeutung der Daten verweisen wir auf die Studie  ['Predicting Group Long Term Disability Recovery and Mortality Rates Using Tree Models'](https://www.soa.org/globalassets/assets/files/research/projects/2017-gltd-recovery-mortality-tree.pdf)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "_kg_hide-input": false,
    "_kg_hide-output": true,
    "execution": {
     "iopub.execute_input": "2020-08-30T09:33:54.972663Z",
     "iopub.status.busy": "2020-08-30T09:33:54.971572Z",
     "iopub.status.idle": "2020-08-30T09:33:55.016994Z",
     "shell.execute_reply": "2020-08-30T09:33:55.009393Z"
    },
    "papermill": {
     "duration": 0.078813,
     "end_time": "2020-08-30T09:33:55.017152",
     "exception": false,
     "start_time": "2020-08-30T09:33:54.938339",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['Unnamed: 0',\n",
      " 'Duration_12_49',\n",
      " 'AgeBand',\n",
      " 'Disability_Category',\n",
      " 'OwnOccToAnyTransition_MOD',\n",
      " 'Integration_with_STD',\n",
      " 'Taxability_Benefits',\n",
      " 'Gender',\n",
      " 'Gross_Indexed_Benefit_Amount',\n",
      " 'Exposures',\n",
      " 'Actual_Recoveries',\n",
      " 'Expected_Recoveries',\n",
      " 'Actual_Deaths',\n",
      " 'Expected_Deaths',\n",
      " 'Actual_Recovery_Rate']\n"
     ]
    }
   ],
   "source": [
    "pprint(list(recovery.columns))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "_kg_hide-output": true,
    "execution": {
     "iopub.execute_input": "2020-08-30T09:33:55.087434Z",
     "iopub.status.busy": "2020-08-30T09:33:55.086662Z",
     "iopub.status.idle": "2020-08-30T09:33:55.096654Z",
     "shell.execute_reply": "2020-08-30T09:33:55.095834Z"
    },
    "papermill": {
     "duration": 0.049752,
     "end_time": "2020-08-30T09:33:55.096812",
     "exception": false,
     "start_time": "2020-08-30T09:33:55.047060",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "array([ 2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15, 16, 17, 18,\n",
      "       19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35,\n",
      "       36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49])\n"
     ]
    }
   ],
   "source": [
    "pprint(recovery['Duration_12_49'].unique())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "_kg_hide-output": true,
    "execution": {
     "iopub.execute_input": "2020-08-30T09:33:55.162918Z",
     "iopub.status.busy": "2020-08-30T09:33:55.161964Z",
     "iopub.status.idle": "2020-08-30T09:33:55.171521Z",
     "shell.execute_reply": "2020-08-30T09:33:55.170743Z"
    },
    "papermill": {
     "duration": 0.044477,
     "end_time": "2020-08-30T09:33:55.171664",
     "exception": false,
     "start_time": "2020-08-30T09:33:55.127187",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "array([20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70])\n"
     ]
    }
   ],
   "source": [
    "pprint(recovery['AgeBand'].unique())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "_kg_hide-output": true,
    "execution": {
     "iopub.execute_input": "2020-08-30T09:33:55.274906Z",
     "iopub.status.busy": "2020-08-30T09:33:55.274142Z",
     "iopub.status.idle": "2020-08-30T09:33:55.304616Z",
     "shell.execute_reply": "2020-08-30T09:33:55.305428Z"
    },
    "papermill": {
     "duration": 0.103902,
     "end_time": "2020-08-30T09:33:55.305625",
     "exception": false,
     "start_time": "2020-08-30T09:33:55.201723",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "array(['Back', 'Cancer', 'Circulatory', 'Digestive',\n",
      "       'Ill-defined and Misc Conditions', 'Injury other than back',\n",
      "       'Maternity', 'Mental and Nervous', 'Nervous System', 'Other',\n",
      "       'Other Musculoskeletal', 'Respiratory', 'Diabetes'], dtype=object)\n"
     ]
    }
   ],
   "source": [
    "pprint(recovery['Disability_Category'].unique())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "_kg_hide-output": true,
    "execution": {
     "iopub.execute_input": "2020-08-30T09:33:55.409718Z",
     "iopub.status.busy": "2020-08-30T09:33:55.408891Z",
     "iopub.status.idle": "2020-08-30T09:33:55.439686Z",
     "shell.execute_reply": "2020-08-30T09:33:55.438768Z"
    },
    "papermill": {
     "duration": 0.103713,
     "end_time": "2020-08-30T09:33:55.439859",
     "exception": false,
     "start_time": "2020-08-30T09:33:55.336146",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "array(['OwnOther', 'Own+1', 'Own+0'], dtype=object)\n"
     ]
    }
   ],
   "source": [
    "pprint(recovery['OwnOccToAnyTransition_MOD'].unique())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "_kg_hide-output": true,
    "execution": {
     "iopub.execute_input": "2020-08-30T09:33:55.562701Z",
     "iopub.status.busy": "2020-08-30T09:33:55.561651Z",
     "iopub.status.idle": "2020-08-30T09:33:55.627570Z",
     "shell.execute_reply": "2020-08-30T09:33:55.628312Z"
    },
    "papermill": {
     "duration": 0.14284,
     "end_time": "2020-08-30T09:33:55.628523",
     "exception": false,
     "start_time": "2020-08-30T09:33:55.485683",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "array(['I: Integrated with ASO of Fully-Insured STD',\n",
      "       'N: Not Integrated with STD', 'U: Unknown'], dtype=object)\n"
     ]
    }
   ],
   "source": [
    "pprint(recovery['Integration_with_STD'].unique())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "_kg_hide-output": true,
    "execution": {
     "iopub.execute_input": "2020-08-30T09:33:55.743038Z",
     "iopub.status.busy": "2020-08-30T09:33:55.742266Z",
     "iopub.status.idle": "2020-08-30T09:33:55.777607Z",
     "shell.execute_reply": "2020-08-30T09:33:55.776904Z"
    },
    "papermill": {
     "duration": 0.110044,
     "end_time": "2020-08-30T09:33:55.777735",
     "exception": false,
     "start_time": "2020-08-30T09:33:55.667691",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "array(['T: 100% Taxable', 'Unknown/Invalid', 'N: Non-Taxable',\n",
      "       'P: Partial Taxability'], dtype=object)\n"
     ]
    }
   ],
   "source": [
    "pprint(recovery['Taxability_Benefits'].unique())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "_kg_hide-output": true,
    "execution": {
     "iopub.execute_input": "2020-08-30T09:33:55.884018Z",
     "iopub.status.busy": "2020-08-30T09:33:55.883014Z",
     "iopub.status.idle": "2020-08-30T09:33:55.909511Z",
     "shell.execute_reply": "2020-08-30T09:33:55.908655Z"
    },
    "papermill": {
     "duration": 0.100677,
     "end_time": "2020-08-30T09:33:55.909673",
     "exception": false,
     "start_time": "2020-08-30T09:33:55.808996",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "array(['2:Female', '1: Male'], dtype=object)\n"
     ]
    }
   ],
   "source": [
    "pprint(recovery['Gender'].unique())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "_kg_hide-output": true,
    "execution": {
     "iopub.execute_input": "2020-08-30T09:33:56.018609Z",
     "iopub.status.busy": "2020-08-30T09:33:56.017775Z",
     "iopub.status.idle": "2020-08-30T09:33:56.057971Z",
     "shell.execute_reply": "2020-08-30T09:33:56.057253Z"
    },
    "papermill": {
     "duration": 0.116042,
     "end_time": "2020-08-30T09:33:56.058124",
     "exception": false,
     "start_time": "2020-08-30T09:33:55.942082",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "array(['1: $ < 1000', '2: $1,000 - $1,999', '3: $2,000 - $2,999',\n",
      "       '4: $3,000 - $3,999', '5: $4,000 - $4,999', '6: $5,000 - $9,999',\n",
      "       'Unknown', '7:$10,000 - $14,999', '9:$20,000 And Over',\n",
      "       '8:$15,000 - $19,999'], dtype=object)\n"
     ]
    }
   ],
   "source": [
    "pprint(recovery['Gross_Indexed_Benefit_Amount'].unique())"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.031376,
     "end_time": "2020-08-30T09:33:56.121177",
     "exception": false,
     "start_time": "2020-08-30T09:33:56.089801",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Um ein Gefühl für die Ausprägungen der Variablen zu bekommen, sind die Häufigkeiten ihres Auftretens nachfolgend in Histogrammen geplottet. \n",
    "In den Histogrammen wird deutlich, dass in etwa ähnlich viele Männer wie Frauen teilgenommen haben. Die am häufigsten auftretende Altergruppen liegt zwischen 50 und 60 Jahren. Insgesamt werden 13 Gründe untersucht, weshalb jemand berufusunfähig geworden ist, wobei am häufigsten die Ursache \"back\" angegeben wurde. \n",
    "Insbesondere erkennt man auch, dass die Tatsächliche Recovery Rate sehr schief verteilt ist mit einer Häufung bei 0. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "_kg_hide-input": true,
    "execution": {
     "iopub.execute_input": "2020-08-30T09:33:56.221110Z",
     "iopub.status.busy": "2020-08-30T09:33:56.220293Z",
     "iopub.status.idle": "2020-08-30T09:34:04.611886Z",
     "shell.execute_reply": "2020-08-30T09:34:04.612584Z"
    },
    "papermill": {
     "duration": 8.459137,
     "end_time": "2020-08-30T09:34:04.612795",
     "exception": false,
     "start_time": "2020-08-30T09:33:56.153658",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 2160x576 with 14 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "diagramm = plt.figure(figsize=(30,8))\n",
    "plt.style.use('ggplot')\n",
    "plt.subplots_adjust(bottom=0.1, right=0.8, top=2)\n",
    "\n",
    "# Duration_12_49\n",
    "dDuration = recovery['Duration_12_49']\n",
    "aDuration = diagramm.add_subplot(5,3,1)\n",
    "plt.xticks(rotation=90)\n",
    "plt.title('Duration_12_49')\n",
    "plt.hist(dDuration, bins=100)\n",
    "\n",
    "# AgeBand\n",
    "dAgeBand = recovery['AgeBand']\n",
    "aAgeBand = diagramm.add_subplot(5,3,2)\n",
    "plt.xticks(rotation=90)\n",
    "plt.title('AgeBand')\n",
    "plt.hist(dAgeBand, bins=100)\n",
    "\n",
    "# Disability_Category\n",
    "dDisCat = recovery['Disability_Category']\n",
    "aDisCat = diagramm.add_subplot(5,3,3)\n",
    "plt.xticks(rotation=90)\n",
    "plt.title('Disability_Category')\n",
    "plt.hist(dDisCat, bins=100)\n",
    "\n",
    "# OwnOccToAnyTransition_MOD\n",
    "dOwnOCC = recovery['OwnOccToAnyTransition_MOD']\n",
    "aOwnOCC = diagramm.add_subplot(5,3,4)\n",
    "plt.xticks(rotation=90)\n",
    "plt.title('OwnOccToAnyTransition_MOD')\n",
    "plt.hist(dOwnOCC, bins=100)\n",
    "\n",
    "# Integration_with_STD\n",
    "dIntSTD = recovery['Integration_with_STD']\n",
    "aIntSTD = diagramm.add_subplot(5,3,5)\n",
    "plt.xticks(rotation=90)\n",
    "plt.title('Integration_with_STD')\n",
    "plt.hist(dIntSTD, bins=100)\n",
    "\n",
    "# Taxability_Benefits\n",
    "dTaxBen = recovery['Taxability_Benefits']\n",
    "aTaxBen = diagramm.add_subplot(5,3,6)\n",
    "plt.xticks(rotation=90)\n",
    "plt.title('Taxability_Benefits')\n",
    "plt.hist(dTaxBen, bins=100)\n",
    "\n",
    "# Gender\n",
    "dGender = recovery['Gender']\n",
    "aGender = diagramm.add_subplot(5,3,7)\n",
    "plt.xticks(rotation=90)\n",
    "plt.title('Gender')\n",
    "plt.hist(dGender, bins=100)\n",
    "\n",
    "# Gross_Indexed_Benefit_Amount\n",
    "dGIBA = recovery['Gross_Indexed_Benefit_Amount']\n",
    "aGIBA = diagramm.add_subplot(5,3,8)\n",
    "plt.xticks(rotation=90)\n",
    "plt.title('Gross_Indexed_Benefit_Amount')\n",
    "plt.hist(dGIBA, bins=100)\n",
    "\n",
    "# Exposures\n",
    "dExposures = recovery['Exposures']\n",
    "aExposures = diagramm.add_subplot(5,3,9)\n",
    "plt.xticks(rotation=90)\n",
    "plt.title('Exposures')\n",
    "plt.hist(dExposures, bins=100)\n",
    "\n",
    "# Actual_Recoveries\n",
    "dActRec = recovery['Actual_Recoveries']\n",
    "aActRec = diagramm.add_subplot(5,3,10)\n",
    "plt.xticks(rotation=90)\n",
    "plt.title('Actual_Recoveries')\n",
    "plt.hist(dActRec, bins=100)\n",
    "\n",
    "# Expected_Recoveries\n",
    "dExpRec = recovery['Expected_Recoveries']\n",
    "aExpRec = diagramm.add_subplot(5,3,11)\n",
    "plt.xticks(rotation=90)\n",
    "plt.title('Expected_Recoveries')\n",
    "plt.hist(dExpRec, bins=100)\n",
    "\n",
    "# Actual_Deaths\n",
    "dActDeaths = recovery['Actual_Deaths']\n",
    "aActDeaths = diagramm.add_subplot(5,3,12)\n",
    "plt.xticks(rotation=90)\n",
    "plt.title('Actual_Deaths')\n",
    "plt.hist(dActDeaths, bins=100)\n",
    "\n",
    "# Expected_Deaths\n",
    "dExpDeaths = recovery['Expected_Deaths']\n",
    "aExpDeaths = diagramm.add_subplot(5,3,13)\n",
    "plt.xticks(rotation=90)\n",
    "plt.title('Expected_Deaths')\n",
    "plt.hist(dExpDeaths, bins=100)\n",
    "\n",
    "# Actual_Recovery_Rate\n",
    "dActRecRate = recovery['Actual_Recovery_Rate']\n",
    "aActRecRate = diagramm.add_subplot(5,3,14)\n",
    "plt.xticks(rotation=90)\n",
    "plt.title('Actual_Recovery_Rate')\n",
    "plt.hist(dActRecRate, bins=100)\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.032022,
     "end_time": "2020-08-30T09:34:04.677842",
     "exception": false,
     "start_time": "2020-08-30T09:34:04.645820",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Analog zum Vorgehen der SOA haben wir die Actual Recovery Rate aufgrund der schiefen Verteilung für die Datenpunkte unter Ausschluss der Personen mit einer Actual Recovery Rate in Höhe von 0.0 zusätzlich geplottet. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:04.803481Z",
     "iopub.status.busy": "2020-08-30T09:34:04.802358Z",
     "iopub.status.idle": "2020-08-30T09:34:05.754495Z",
     "shell.execute_reply": "2020-08-30T09:34:05.755043Z"
    },
    "papermill": {
     "duration": 1.045086,
     "end_time": "2020-08-30T09:34:05.755198",
     "exception": false,
     "start_time": "2020-08-30T09:34:04.710112",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "diagramm = plt.figure(figsize=(14,4))\n",
    "plt.style.use('ggplot')\n",
    "# mit 0-Werte\n",
    "dActRec = recovery['Actual_Recovery_Rate']\n",
    "aActRec = diagramm.add_subplot(1,2,1)\n",
    "plt.hist(dActRec, bins=100)\n",
    "\n",
    "# ohne 0-Werte\n",
    "np.count_nonzero(recovery['Actual_Recovery_Rate']==0)/ len(recovery['Actual_Recovery_Rate'])\n",
    "#76%  of the values of Actual_Recovery_Rate are equal to  0 . A histogram of the actual recovery rate is presented next:\n",
    "dActRecoverRatePos = [x for x in recovery['Actual_Recovery_Rate'] if x != 0]\n",
    "aActRecoverRatePos = diagramm.add_subplot(1,2,2)\n",
    "plt.hist(dActRecoverRatePos, bins=100)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.032027,
     "end_time": "2020-08-30T09:34:05.819669",
     "exception": false,
     "start_time": "2020-08-30T09:34:05.787642",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "<a id=\"VonRNachPython\"></a>\n",
    "# Übersetzung von R nach Python \n",
    "\n",
    "Im folgenden werden die Ergebnisse der SOA in der Programmiersprache \"Python\" reproduziert. Zunächst werden die Trainings- & Testdaten eingelesen. Danach folgt die Analyse der folgenden Machine-Learning Verfahren:\n",
    "* GLM - Generalized Linear Models\n",
    "* CART - Classification and Regression Trees\n",
    "* BAGGING - Bootstrap Aggregation\n",
    "* RF - Random forests\n",
    "* XGB - Gradient Boosting\n",
    "* BART - Bayesian Additive Regressions Trees\n",
    "* MARS - Multivariate Adaptive Regression Splines\n",
    "\n",
    "Auf die genaue Beschreibung der reproduzierten Machine Learning Methoden verzichten wir aus Gründen Übersichtlichkeit an dieser Stelle und verweisen auf den Bericht der SOA.\n",
    "\n",
    "Zunächst definieren wir für die einzelnen Modelle noch eine Methode, die uns die Ergebnisse des jeweiligen Modells anzeigen kann."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:05.894980Z",
     "iopub.status.busy": "2020-08-30T09:34:05.894105Z",
     "iopub.status.idle": "2020-08-30T09:34:05.896904Z",
     "shell.execute_reply": "2020-08-30T09:34:05.897450Z"
    },
    "papermill": {
     "duration": 0.045402,
     "end_time": "2020-08-30T09:34:05.897613",
     "exception": false,
     "start_time": "2020-08-30T09:34:05.852211",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "def output(model, python_mse, r_mse):\n",
    "    try:\n",
    "        python_mse\n",
    "        r_mse\n",
    "    except NameError:\n",
    "        print(model + \": Nicht erfolgreich ausgeführt. Variablen unbekannt.\")\n",
    "    else:\n",
    "        print(model + \": Erfolgreich ausgeführt\")\n",
    "        print(model + \": Ergebnis mse = \" + str(python_mse))\n",
    "        if r_mse != None:\n",
    "            print(model + \": Abwechung zum entsprechenden Modell in 'R':\" + str(python_mse - r_mse))\n",
    "            if python_mse - r_mse < 0:\n",
    "                print(model + \": Das Modell in 'Python' ist besser!\")\n",
    "            elif python_mse == r_mse:\n",
    "                print(model + \": Die Modelle sind gleichwertig!\")\n",
    "            else:\n",
    "                print(model + \": Das Modell in 'R' ist besser!\")"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.032788,
     "end_time": "2020-08-30T09:34:05.963135",
     "exception": false,
     "start_time": "2020-08-30T09:34:05.930347",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Da wir den Großteil der Elemente dieser Arbeit in Kaggle entwickelt haben, definieren wir um die Laufzeit zu verkürzen einen boolschen Wert 'mitBerechnung'. Weißt dieser den Wert 'True' auf, so werden alle Berechnungen innerhalb dieses Notebooks ausgeführt, ist der Wert hingegen 'False', so greifen wir auf Werte zurück, die wir in einer der vorherigen Versionen des Notebooks erzeugt haben. Natürlich werden nicht alle Ergebnisse in jeder Ausführung auch wieder exakt getroffen. Die Variationen erschienen uns allerdings vernachlässigbar."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:06.034790Z",
     "iopub.status.busy": "2020-08-30T09:34:06.033710Z",
     "iopub.status.idle": "2020-08-30T09:34:06.036928Z",
     "shell.execute_reply": "2020-08-30T09:34:06.036284Z"
    },
    "papermill": {
     "duration": 0.040645,
     "end_time": "2020-08-30T09:34:06.037072",
     "exception": false,
     "start_time": "2020-08-30T09:34:05.996427",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "mitBerechnung = False "
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.032352,
     "end_time": "2020-08-30T09:34:06.102630",
     "exception": false,
     "start_time": "2020-08-30T09:34:06.070278",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "<a id=\"DefinitionTrainingsTestdaten\"></a>\n",
    "# Definition der Trainings- bzw. Testdaten"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.032622,
     "end_time": "2020-08-30T09:34:06.168155",
     "exception": false,
     "start_time": "2020-08-30T09:34:06.135533",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Zielgröße ist die Variable \"Actual_Recovery_Rate\". Die untersuchten Einflussgrößen sind die sieben Variablen:\n",
    "1. Duration \n",
    "2. Age \n",
    "3. Gender \n",
    "4. Disability Category \n",
    "5. Own to Any Occupation Transition  \n",
    "6. Gross Indexed Benefit Amount \n",
    "7. Integration with STD \n",
    "\n",
    "Zum Trainieren der Machine-Learning-Methoden werden Trainingsdaten benötigt. Zusätzlich dazu werden Testdaten ermittelt anhand derer die Performance der verschiedenen Methoden gemessen wird.\n",
    "Um die Ergebnisse vergleichbar zu machen, haben wir den Trainingsdatensatz der SOA aus R exportiert und hier wieder importiert. Alternativ können an dieser Stelle neue Trainingsdaten erzeugt werden, indem zufällig 70% der Datenpunkte als Trainingsdatensatz ausgewählt und die verbleibenden 30% als Testdatensatz verwendet werden."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:06.249815Z",
     "iopub.status.busy": "2020-08-30T09:34:06.246739Z",
     "iopub.status.idle": "2020-08-30T09:34:09.023263Z",
     "shell.execute_reply": "2020-08-30T09:34:09.022619Z"
    },
    "papermill": {
     "duration": 2.822706,
     "end_time": "2020-08-30T09:34:09.023386",
     "exception": false,
     "start_time": "2020-08-30T09:34:06.200680",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Trainings- und Testdaten erfolgreich generiert.\n"
     ]
    }
   ],
   "source": [
    "# Generierung einer neuen Testmenge mit 1, lesen der CSV-Dateien mit 0\n",
    "x = 0\n",
    "if x == 1:\n",
    "    recoveryMod = recovery.loc[:,\n",
    "        ('Actual_Recovery_Rate', 'AgeBand', 'Duration_12_49', 'Gross_Indexed_Benefit_Amount', 'Disability_Category',\n",
    "         'OwnOccToAnyTransition_MOD', 'Integration_with_STD', 'Gender')]\n",
    "    # mischen der Daten\n",
    "    random.seed(1)\n",
    "    index= list(range(0,recoveryMod.shape[0]))\n",
    "    random.shuffle(index)\n",
    "    # 70% der Daten werden zum Trainieren verwendet:\n",
    "    train_sample = math.floor(0.7 * len(recoveryMod))\n",
    "    datatrain = recoveryMod.iloc[index[:train_sample], :]\n",
    "    datatest = recoveryMod.iloc[index[train_sample:], :]\n",
    "else: \n",
    "    datatrain_full = pd.read_csv('../input/gltdtestdata/dataTrain.csv')\n",
    "    datatrain = datatrain_full[['Actual_Recovery_Rate','AgeBand', 'Duration_12_49', 'Gross_Indexed_Benefit_Amount', 'Disability_Category', 'OwnOccToAnyTransition_MOD', 'Integration_with_STD', 'Gender']]\n",
    "    datatest_full = pd.read_csv('../input/gltdtestdata/dataTest.csv')\n",
    "    datatest = datatest_full[['Actual_Recovery_Rate','AgeBand', 'Duration_12_49', 'Gross_Indexed_Benefit_Amount', 'Disability_Category', 'OwnOccToAnyTransition_MOD', 'Integration_with_STD', 'Gender']]\n",
    "    \n",
    "x_train = datatrain.loc[:, datatrain.columns!='Actual_Recovery_Rate']\n",
    "x_train_ohe = pd.get_dummies(x_train)\n",
    "y_train = datatrain.loc[:, datatrain.columns=='Actual_Recovery_Rate']    \n",
    "x_test = datatest.loc[:, datatest.columns!='Actual_Recovery_Rate']\n",
    "x_test_ohe = pd.get_dummies(x_test)\n",
    "y_test = datatest.loc[:, datatrain.columns=='Actual_Recovery_Rate']     \n",
    "\n",
    "print(\"Trainings- und Testdaten erfolgreich generiert.\")"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.032708,
     "end_time": "2020-08-30T09:34:09.089211",
     "exception": false,
     "start_time": "2020-08-30T09:34:09.056503",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "<a id=\"GLM\"></a>\n",
    "## GLM - Generalized Linear Models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:09.166303Z",
     "iopub.status.busy": "2020-08-30T09:34:09.165379Z",
     "iopub.status.idle": "2020-08-30T09:34:09.178443Z",
     "shell.execute_reply": "2020-08-30T09:34:09.177776Z"
    },
    "papermill": {
     "duration": 0.056672,
     "end_time": "2020-08-30T09:34:09.178570",
     "exception": false,
     "start_time": "2020-08-30T09:34:09.121898",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GLM 1: Erfolgreich ausgeführt\n",
      "GLM 1: Ergebnis mse = 0.013338525987274875\n",
      "GLM 1: Abwechung zum entsprechenden Modell in 'R':-4.01272512440265e-09\n",
      "GLM 1: Das Modell in 'Python' ist besser!\n"
     ]
    }
   ],
   "source": [
    "# GLM 1\n",
    "if mitBerechnung:\n",
    "    # Zeitmessung starten\n",
    "    start_time = datetime.datetime.now()\n",
    "    \n",
    "    formula = 'Actual_Recovery_Rate ~ AgeBand + Duration_12_49 + Gross_Indexed_Benefit_Amount + Disability_Category + OwnOccToAnyTransition_MOD + Integration_with_STD + Gender'\n",
    "    model = smf.glm(formula = formula, data=datatrain, family= sm.families.Gaussian())  #sm.families.Gaussian(sm.families.links.identity)) \n",
    "    result = model.fit()\n",
    "    ypred = result.predict(datatest)\n",
    "    mse_glm1 = ((ypred - datatest['Actual_Recovery_Rate'])**2).mean(axis=None) \n",
    "    \n",
    "    # Zeitmessung beenden\n",
    "    end_time = datetime.datetime.now()\n",
    "\n",
    "    # verbrauchte Zeit berechnen\n",
    "    elapsed_time = end_time - start_time\n",
    "    t_glm1 = elapsed_time.total_seconds()\n",
    "else:\n",
    "    mse_glm1 = 0.013338525987274875\n",
    "    t_glm1 = 16.404506\n",
    "\n",
    "# Vergleichswerte aus R\n",
    "r_mse_glm1 = 0.01333853\n",
    "r_t_glm1 = 3.230486\n",
    "\n",
    "# Output\n",
    "output(\"GLM 1\", mse_glm1, r_mse_glm1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:09.267131Z",
     "iopub.status.busy": "2020-08-30T09:34:09.265931Z",
     "iopub.status.idle": "2020-08-30T09:34:09.279060Z",
     "shell.execute_reply": "2020-08-30T09:34:09.279670Z"
    },
    "papermill": {
     "duration": 0.068001,
     "end_time": "2020-08-30T09:34:09.279820",
     "exception": false,
     "start_time": "2020-08-30T09:34:09.211819",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GLM 2: Erfolgreich ausgeführt\n",
      "GLM 2: Ergebnis mse = 0.014760995460538455\n",
      "GLM 2: Abwechung zum entsprechenden Modell in 'R':-4.539461544508794e-09\n",
      "GLM 2: Das Modell in 'Python' ist besser!\n"
     ]
    }
   ],
   "source": [
    "# GLM 2\n",
    "if mitBerechnung:\n",
    "    # Zeitmessung starten\n",
    "    start_time = datetime.datetime.now()\n",
    "    \n",
    "    formula_model1 = \"\"\"Actual_Recovery_Rate2 ~ Disability_Category\n",
    "                            + AgeBand + Duration_12_49 + OwnOccToAnyTransition_MOD\n",
    "                            + Integration_with_STD + Gender + Gross_Indexed_Benefit_Amount\"\"\"\n",
    "    \n",
    "    data_mod = datatrain\n",
    "    data_mod.loc[:,'Actual_Recovery_Rate2'] = np.where(datatrain.loc[:,'Actual_Recovery_Rate'] > 0, 1, 0)\n",
    "    glm_model1 = smf.glm(formula=formula_model1, data=data_mod, family=sm.families.Binomial())\n",
    "    result1 = glm_model1.fit()\n",
    "    \n",
    "    prediction1 = result1.predict(datatest)\n",
    "    glm_prediction = np.where(prediction1 > 0.5, 1, 0)\n",
    "    \n",
    "    data_mod_log  = datatrain[datatrain['Actual_Recovery_Rate'] != 0]\n",
    "    data_mod_log.loc[:]['Actual_Recovery_Rate'] = np.log(data_mod_log['Actual_Recovery_Rate'])\n",
    "    formula_model2 = 'Actual_Recovery_Rate ~ Disability_Category + AgeBand + Duration_12_49 + OwnOccToAnyTransition_MOD + Integration_with_STD + Gender + Gross_Indexed_Benefit_Amount'\n",
    "    glm_model2 = smf.glm(formula=formula_model2, data=data_mod_log,family=sm.families.Gaussian())\n",
    "    \n",
    "    result2 = glm_model2.fit()\n",
    "\n",
    "    indices = [i for i, v in enumerate(glm_prediction) if glm_prediction[i] > 0]\n",
    "    testset = datatest.loc[:,datatest.columns != 'Actual_Recovery_Rate']\n",
    "    testset = testset.loc[indices]\n",
    "    \n",
    "    prediction = np.where(prediction1==1, 1.0, 0.0)\n",
    "\n",
    "    prediction2 = result2.predict(testset)\n",
    "    glm_prediction = np.exp(np.array(prediction2))\n",
    "    for i, v in enumerate(indices):\n",
    "        prediction[v] =  glm_prediction[i]\n",
    "\n",
    "    mse_glm2 = ((prediction - datatest['Actual_Recovery_Rate']) ** 2).mean(axis=None)\n",
    "    \n",
    "    # Zeitmessung beenden\n",
    "    end_time = datetime.datetime.now()\n",
    "    \n",
    "    # verbrauchte Zeit berechnen\n",
    "    elapsed_time = end_time - start_time\n",
    "    t_glm2 = elapsed_time.total_seconds()\n",
    "else:\n",
    "    mse_glm2 = 0.014760995460538455\n",
    "    t_glm2 = 21.370961\n",
    "\n",
    "# Vergleichswerte aus R\n",
    "r_mse_glm2 = 0.014761\n",
    "r_t_glm2 = 9.107313\n",
    "\n",
    "# Output\n",
    "output(\"GLM 2\", mse_glm2, r_mse_glm2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:09.357594Z",
     "iopub.status.busy": "2020-08-30T09:34:09.355662Z",
     "iopub.status.idle": "2020-08-30T09:34:09.370251Z",
     "shell.execute_reply": "2020-08-30T09:34:09.369381Z"
    },
    "papermill": {
     "duration": 0.056817,
     "end_time": "2020-08-30T09:34:09.370387",
     "exception": false,
     "start_time": "2020-08-30T09:34:09.313570",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GLM 3: Erfolgreich ausgeführt\n",
      "GLM 3: Ergebnis mse = 0.012837709032298566\n",
      "GLM 3: Abwechung zum entsprechenden Modell in 'R':-9.67701434609336e-10\n",
      "GLM 3: Das Modell in 'Python' ist besser!\n"
     ]
    }
   ],
   "source": [
    "# GLM 3\n",
    "if mitBerechnung:\n",
    "    # Zeitmessung starten\n",
    "    start_time = datetime.datetime.now()\n",
    "    \n",
    "    formula_model3 = 'Actual_Recovery_Rate ~ (Disability_Category + AgeBand + Duration_12_49 + OwnOccToAnyTransition_MOD + Integration_with_STD + Gender + Gross_Indexed_Benefit_Amount)*(Disability_Category + AgeBand + Duration_12_49 + OwnOccToAnyTransition_MOD + Integration_with_STD + Gender + Gross_Indexed_Benefit_Amount)'\n",
    "    \n",
    "    glm_model3 = smf.glm(formula = formula_model3, data = datatrain, family = sm.families.Gaussian())\n",
    "    \n",
    "    glm_result3 = glm_model3.fit()\n",
    "    glm_predictions3 = glm_result3.predict(datatest)\n",
    "    \n",
    "    mse_glm3 = ((glm_predictions3 - datatest['Actual_Recovery_Rate'])**2).mean(axis=None) \n",
    "    \n",
    "    # Zeitmessung beenden\n",
    "    end_time = datetime.datetime.now()\n",
    "    \n",
    "    # verbrauchte Zeit berechnen\n",
    "    elapsed_time = end_time - start_time\n",
    "    t_glm3 = elapsed_time.total_seconds()\n",
    "else:\n",
    "    mse_glm3 = 0.012837709032298566\n",
    "    t_glm3 = 85.949018 \n",
    "\n",
    "# Vergleichswerte aus R\n",
    "r_mse_glm3 = 0.01283771\n",
    "r_t_glm3 = 94.1341\n",
    "\n",
    "# Output    \n",
    "output(\"GLM 3\", mse_glm3, r_mse_glm3)"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.033715,
     "end_time": "2020-08-30T09:34:09.438207",
     "exception": false,
     "start_time": "2020-08-30T09:34:09.404492",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "<a id=\"CART\"></a>\n",
    "## CART - Classification and Regression Trees"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:09.518554Z",
     "iopub.status.busy": "2020-08-30T09:34:09.517318Z",
     "iopub.status.idle": "2020-08-30T09:34:09.531129Z",
     "shell.execute_reply": "2020-08-30T09:34:09.531726Z"
    },
    "papermill": {
     "duration": 0.059363,
     "end_time": "2020-08-30T09:34:09.531880",
     "exception": false,
     "start_time": "2020-08-30T09:34:09.472517",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CART: Erfolgreich ausgeführt\n",
      "CART: Ergebnis mse = 0.012192802558206421\n",
      "CART: Abwechung zum entsprechenden Modell in 'R':5.5202558206421046e-05\n",
      "CART: Das Modell in 'R' ist besser!\n"
     ]
    }
   ],
   "source": [
    "# CART\n",
    "if mitBerechnung:\n",
    "    # Zeitmessung starten\n",
    "    start_time = datetime.datetime.now()\n",
    "    \n",
    "    regressor = tree.DecisionTreeRegressor(max_depth = 15, min_samples_leaf = 110, min_impurity_decrease = 0.0000001)\n",
    "    regressor.fit(x_train_ohe.values, y_train.values.ravel())\n",
    "    \n",
    "    predictions = regressor.predict(x_test_ohe.values)\n",
    "    \n",
    "    mse_cart = ((predictions - datatest['Actual_Recovery_Rate'])**2).mean(axis=None)\n",
    "    \n",
    "    # Zeitmessung beenden\n",
    "    end_time = datetime.datetime.now()\n",
    "    \n",
    "    # verbrauchte Zeit berechnen\n",
    "    elapsed_time = end_time - start_time\n",
    "    t_cart = elapsed_time.total_seconds()\n",
    "else :\n",
    "    mse_cart = 0.012192802558206421\n",
    "    t_cart = 3.261407\n",
    "    \n",
    "# Vergleichswerte aus R\n",
    "r_mse_cart = 0.0121376\n",
    "r_t_cart = 90.17855\n",
    "\n",
    "# Output\n",
    "output(\"CART\", mse_cart, r_mse_cart)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:09.608969Z",
     "iopub.status.busy": "2020-08-30T09:34:09.608048Z",
     "iopub.status.idle": "2020-08-30T09:34:09.611391Z",
     "shell.execute_reply": "2020-08-30T09:34:09.610715Z"
    },
    "papermill": {
     "duration": 0.044625,
     "end_time": "2020-08-30T09:34:09.611521",
     "exception": false,
     "start_time": "2020-08-30T09:34:09.566896",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "if mitBerechnung:\n",
    "    dot_data = tree.export_graphviz(regressor, out_file=None,  \n",
    "                filled=True, rounded=True,\n",
    "                special_characters=True)\n",
    "    \n",
    "    graph = graphviz.Source(dot_data)\n",
    "    graph"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.037938,
     "end_time": "2020-08-30T09:34:09.685608",
     "exception": false,
     "start_time": "2020-08-30T09:34:09.647670",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "<a id=\"Bagging\"></a>\n",
    "## BAGGING - Bootstrap Aggregation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:09.768589Z",
     "iopub.status.busy": "2020-08-30T09:34:09.764703Z",
     "iopub.status.idle": "2020-08-30T09:34:09.780797Z",
     "shell.execute_reply": "2020-08-30T09:34:09.781449Z"
    },
    "papermill": {
     "duration": 0.06103,
     "end_time": "2020-08-30T09:34:09.781611",
     "exception": false,
     "start_time": "2020-08-30T09:34:09.720581",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "BAG: Erfolgreich ausgeführt\n",
      "BAG: Ergebnis mse = 0.012138850493853439\n",
      "BAG: Abwechung zum entsprechenden Modell in 'R':3.7000493853437885e-05\n",
      "BAG: Das Modell in 'R' ist besser!\n"
     ]
    }
   ],
   "source": [
    "# bagging\n",
    "if mitBerechnung:\n",
    "    # Zeitmessung starten\n",
    "    start_time = datetime.datetime.now()\n",
    "    tree_sequence = range(50,500,5) \n",
    "    MSE_bag = []\n",
    "    \n",
    "    for i in tree_sequence:\n",
    "        n_estimators = i\n",
    "        regressor = RandomForestRegressor(n_estimators, min_samples_split = 150, max_features=3)\n",
    "        regressor.fit(x_train_ohe.values, y_train.values.ravel())\n",
    "        predictions = regressor.predict(x_test_ohe.values)\n",
    "        MSE_bag.append(((predictions - datatest['Actual_Recovery_Rate'])**2).mean(axis=None))\n",
    "    \n",
    "    mse_bag = np.mean(MSE_bag)\n",
    "    \n",
    "    # Zeitmessung beenden\n",
    "    end_time = datetime.datetime.now()\n",
    "    \n",
    "    # verbrauchte Zeit berechnen\n",
    "    elapsed_time = end_time - start_time\n",
    "    t_bag = elapsed_time.total_seconds()\n",
    "else:\n",
    "    mse_bag = 0.012138850493853439\n",
    "    t_bag = 14259.258127\n",
    "\n",
    "# Vergleichswerte aus R\n",
    "r_mse_bag = 0.01210185\n",
    "r_t_bag = 19205.32\n",
    "\n",
    "# Output\n",
    "output(\"BAG\", mse_bag, r_mse_bag)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:09.862147Z",
     "iopub.status.busy": "2020-08-30T09:34:09.860903Z",
     "iopub.status.idle": "2020-08-30T09:34:09.864694Z",
     "shell.execute_reply": "2020-08-30T09:34:09.864053Z"
    },
    "papermill": {
     "duration": 0.047812,
     "end_time": "2020-08-30T09:34:09.864818",
     "exception": false,
     "start_time": "2020-08-30T09:34:09.817006",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "if mitBerechnung:\n",
    "    x = pd.to_numeric(tree_sequence)\n",
    "    y = pd.to_numeric(MSE_bag)\n",
    "    \n",
    "    diagramm = plt.figure(figsize=(6,6))\n",
    "    plt.style.use('ggplot')\n",
    "\n",
    "    plt.plot(x,y,'o', label=\"MSE\")\n",
    "    plt.title('Übersicht der MSE')\n",
    "    plt.ylabel('MSE')\n",
    "    plt.xlabel('Anzahl der Bäume')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.035322,
     "end_time": "2020-08-30T09:34:09.935419",
     "exception": false,
     "start_time": "2020-08-30T09:34:09.900097",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "<a id=\"Randomforests\"></a>\n",
    "## RF - Random forests"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:10.016609Z",
     "iopub.status.busy": "2020-08-30T09:34:10.014392Z",
     "iopub.status.idle": "2020-08-30T09:34:10.029862Z",
     "shell.execute_reply": "2020-08-30T09:34:10.030503Z"
    },
    "papermill": {
     "duration": 0.06016,
     "end_time": "2020-08-30T09:34:10.030661",
     "exception": false,
     "start_time": "2020-08-30T09:34:09.970501",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "RF: Erfolgreich ausgeführt\n",
      "RF: Ergebnis mse = 0.012137927171212553\n",
      "RF: Abwechung zum entsprechenden Modell in 'R':-0.0002480328287874471\n",
      "RF: Das Modell in 'Python' ist besser!\n"
     ]
    }
   ],
   "source": [
    "# RF\n",
    "if mitBerechnung:\n",
    "    # Zeitmessung starten\n",
    "    start_time = datetime.datetime.now()\n",
    "    \n",
    "    regressor = RandomForestRegressor(n_estimators=250, min_samples_split = 150, max_features=3)\n",
    "    regressor.fit(x_train_ohe.values, y_train.values.ravel())\n",
    "    \n",
    "    predictions = regressor.predict(x_test_ohe.values)\n",
    "    \n",
    "    mse_rf = ((predictions - datatest['Actual_Recovery_Rate'])**2).mean(axis=None)\n",
    "    \n",
    "    # Zeitmessung beenden\n",
    "    end_time = datetime.datetime.now()\n",
    "    \n",
    "    # verbrauchte Zeit berechnen\n",
    "    elapsed_time = end_time - start_time\n",
    "    t_rf = elapsed_time.total_seconds()\n",
    "else:\n",
    "    mse_rf = 0.012137927171212553\n",
    "    t_rf = 153.292926\n",
    "    \n",
    "# Vergleichswerte aus R\n",
    "r_mse_rf = 0.01238596\n",
    "r_t_rf = 91.52749\n",
    "\n",
    "# Output\n",
    "output(\"RF\", mse_rf, r_mse_rf)"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.035619,
     "end_time": "2020-08-30T09:34:10.102567",
     "exception": false,
     "start_time": "2020-08-30T09:34:10.066948",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "<a id=\"GradientBoosting\"></a>\n",
    "## XGB - Gradient Boosting\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:10.188106Z",
     "iopub.status.busy": "2020-08-30T09:34:10.186762Z",
     "iopub.status.idle": "2020-08-30T09:34:10.201337Z",
     "shell.execute_reply": "2020-08-30T09:34:10.200658Z"
    },
    "papermill": {
     "duration": 0.062681,
     "end_time": "2020-08-30T09:34:10.201472",
     "exception": false,
     "start_time": "2020-08-30T09:34:10.138791",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "XGB: Erfolgreich ausgeführt\n",
      "XGB: Ergebnis mse = 0.012083109223432047\n",
      "XGB: Abwechung zum entsprechenden Modell in 'R':-3.1550776567952285e-05\n",
      "XGB: Das Modell in 'Python' ist besser!\n"
     ]
    }
   ],
   "source": [
    "# GradBoost\n",
    "if mitBerechnung:\n",
    "    # Zeitmessung starten\n",
    "    start_time = datetime.datetime.now()      \n",
    "    \n",
    "    xgbtrain = xgb.DMatrix(x_train_ohe.values, label=y_train)\n",
    "    xgbtest = xgb.DMatrix(x_test_ohe.values, label=y_test)\n",
    "    \n",
    "    param = {'max_depth': 4,  'eta': 0.3}\n",
    "    num_round = 1000  # the number of training iterations\n",
    "    \n",
    "    xgb_result = xgb.train(param, xgbtrain, num_round)\n",
    "    \n",
    "    xgb_predictions = xgb_result.predict(xgbtest)\n",
    "    \n",
    "    mse_xgb = ((xgb_predictions - datatest['Actual_Recovery_Rate'])**2).mean(axis=None)\n",
    "    \n",
    "    # Zeitmessung beenden\n",
    "    end_time = datetime.datetime.now()\n",
    "    \n",
    "    # verbrauchte Zeit berechnen\n",
    "    elapsed_time = end_time - start_time\n",
    "    t_xgb = elapsed_time.total_seconds()\n",
    "else:\n",
    "    mse_xgb = 0.012083109223432047\n",
    "    t_xgb = 306.127362\n",
    "    \n",
    "# Vergleichswerte aus R\n",
    "r_mse_xgb = 0.01211466\n",
    "r_t_xgb = 443.8047\n",
    "\n",
    "# Output\n",
    "output(\"XGB\", mse_xgb, r_mse_xgb)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:10.281507Z",
     "iopub.status.busy": "2020-08-30T09:34:10.280550Z",
     "iopub.status.idle": "2020-08-30T09:34:10.282912Z",
     "shell.execute_reply": "2020-08-30T09:34:10.283477Z"
    },
    "papermill": {
     "duration": 0.045264,
     "end_time": "2020-08-30T09:34:10.283648",
     "exception": false,
     "start_time": "2020-08-30T09:34:10.238384",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "if mitBerechnung:\n",
    "    plot_tree(xgb_result, rankdir='LR')\n",
    "    fig = plt.gcf()\n",
    "    fig.set_size_inches(30, 15)"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.036221,
     "end_time": "2020-08-30T09:34:10.356168",
     "exception": false,
     "start_time": "2020-08-30T09:34:10.319947",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "<a id=\"BART\"></a>\n",
    "## BART - Bayesian Additive Regressions Trees\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:10.441659Z",
     "iopub.status.busy": "2020-08-30T09:34:10.440538Z",
     "iopub.status.idle": "2020-08-30T09:34:10.454294Z",
     "shell.execute_reply": "2020-08-30T09:34:10.453627Z"
    },
    "papermill": {
     "duration": 0.061825,
     "end_time": "2020-08-30T09:34:10.454421",
     "exception": false,
     "start_time": "2020-08-30T09:34:10.392596",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "BART: Erfolgreich ausgeführt\n",
      "BART: Ergebnis mse = 0.012872875439113165\n",
      "BART: Abwechung zum entsprechenden Modell in 'R':0.0007708254391131653\n",
      "BART: Das Modell in 'R' ist besser!\n"
     ]
    }
   ],
   "source": [
    "# BART\n",
    "if mitBerechnung:\n",
    "    # Zeitmessung starten\n",
    "    start_time = datetime.datetime.now()\n",
    "    \n",
    "    bart_model = SklearnModel(sublist=None,\n",
    "                     n_trees=5,\n",
    "                     n_chains=4,\n",
    "                     n_samples=50,\n",
    "                     n_burn=200,\n",
    "                     thin=0.1,\n",
    "                     n_jobs=1)   \n",
    "\n",
    "    bart_result = bart_model.fit(x_train_ohe.values, y_train.values.ravel())\n",
    "    \n",
    "    bart_predictions = bart_model.predict(x_test_ohe.values)\n",
    "    \n",
    "    mse_bart = ((bart_predictions - datatest['Actual_Recovery_Rate'])**2).mean(axis=None)\n",
    "    \n",
    "    # Zeitmessung beenden\n",
    "    end_time = datetime.datetime.now()\n",
    "    \n",
    "    # verbrauchte Zeit berechnen\n",
    "    elapsed_time = end_time - start_time\n",
    "    t_bart = elapsed_time.total_seconds()\n",
    "else:\n",
    "    mse_bart = 0.012872875439113165\n",
    "    t_bart = 591.842287\n",
    "    \n",
    "# Vergleichswerte aus R\n",
    "r_mse_bart = 0.01210205\n",
    "r_t_bart = 7284.637\n",
    "\n",
    "# Output  \n",
    "output(\"BART\", mse_bart, r_mse_bart)"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.036328,
     "end_time": "2020-08-30T09:34:10.527255",
     "exception": false,
     "start_time": "2020-08-30T09:34:10.490927",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "<a id=\"MARS\"></a>\n",
    "## MARS - Multivariate Adaptive Regression Splines"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:10.610168Z",
     "iopub.status.busy": "2020-08-30T09:34:10.609159Z",
     "iopub.status.idle": "2020-08-30T09:34:10.621999Z",
     "shell.execute_reply": "2020-08-30T09:34:10.622688Z"
    },
    "papermill": {
     "duration": 0.058954,
     "end_time": "2020-08-30T09:34:10.622844",
     "exception": false,
     "start_time": "2020-08-30T09:34:10.563890",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MARS: Erfolgreich ausgeführt\n",
      "MARS: Ergebnis mse = 0.013400830351710476\n",
      "MARS: Abwechung zum entsprechenden Modell in 'R':0.0009757203517104761\n",
      "MARS: Das Modell in 'R' ist besser!\n"
     ]
    }
   ],
   "source": [
    "# MARS\n",
    "if mitBerechnung:\n",
    "    # Zeitmessung starten\n",
    "    start_time = datetime.datetime.now()\n",
    "    \n",
    "    mars_model = pyearth.Earth()\n",
    "    \n",
    "    mars_result = mars_model.fit(x_train_ohe,y_train)\n",
    "    \n",
    "    mars_prediction = mars_result.predict(x_test_ohe)\n",
    "    \n",
    "    mse_mars = ((mars_prediction - datatest['Actual_Recovery_Rate'])**2).mean(axis=None)\n",
    "    \n",
    "    # Zeitmessung beenden\n",
    "    end_time = datetime.datetime.now()\n",
    "    \n",
    "    # verbrauchte Zeit berechnen\n",
    "    elapsed_time = end_time - start_time\n",
    "    t_mars = elapsed_time.total_seconds()\n",
    "else:\n",
    "    mse_mars = 0.013400830351710476\n",
    "    t_mars = 32.620724\n",
    "\n",
    "# Vergleichswerte aus R\n",
    "r_mse_mars = 0.01242511\n",
    "r_t_mars = 763.7243\n",
    "\n",
    "# Output\n",
    "output(\"MARS\", mse_mars, r_mse_mars)"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.048095,
     "end_time": "2020-08-30T09:34:10.709983",
     "exception": false,
     "start_time": "2020-08-30T09:34:10.661888",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "<a id=\"Fazit\"></a>\n",
    "# Fazit\n",
    "\n",
    "Bevor wir ein Fazit ziehen, fassen wir zunächst die Ergebnisse der vorangegangenen Modelle in einer übersichtlichen Art und Weise zusammen. Die Vergleichswerte der Sprache 'R' erhalten wir dabei aus dem Jupyter-Notebook zu der Studie 'Machine-Learning Methods for Insurance Applications', welches ebenfalls frei zugänglich ist. Auf Wunsch können wir die Version, die wir zum Erstellen der Ergebniswerte verwendet haben, gerne zur Verfügung stellen."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:10.820884Z",
     "iopub.status.busy": "2020-08-30T09:34:10.819968Z",
     "iopub.status.idle": "2020-08-30T09:34:10.840357Z",
     "shell.execute_reply": "2020-08-30T09:34:10.839592Z"
    },
    "papermill": {
     "duration": 0.083204,
     "end_time": "2020-08-30T09:34:10.840482",
     "exception": false,
     "start_time": "2020-08-30T09:34:10.757278",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "'Die Ergebnisse in Python:'\n",
      "  Methode       MSE          Zeit Einheit\n",
      "0    GLM1  0.013339     16.404506     sec\n",
      "1    GLM2  0.014761     21.370961     sec\n",
      "2    GLM3  0.012838     85.949018     sec\n",
      "3    CART  0.012193      3.261407     sec\n",
      "4     BAG  0.012139  14259.258127     sec\n",
      "5      RF  0.012138    153.292926     sec\n",
      "6     XGB  0.012083    306.127362     sec\n",
      "7    BART  0.012873    591.842287     sec\n",
      "8    MARS  0.013401     32.620724     sec\n",
      "'Die Ergebnisse in R:'\n",
      "  Methode       MSE          Zeit Einheit\n",
      "0    GLM1  0.013339      3.230486     sec\n",
      "1    GLM2  0.014761      9.107313     sec\n",
      "2    GLM3  0.012838     94.134100     sec\n",
      "3    CART  0.012138     90.178550     sec\n",
      "4     BAG  0.012102  19205.320000     sec\n",
      "5      RF  0.012386     91.527490     sec\n",
      "6     XGB  0.012115    443.804700     sec\n",
      "7    BART  0.012102   7284.637000     sec\n",
      "8    MARS  0.012425    763.724300     sec\n"
     ]
    }
   ],
   "source": [
    "# Zusammenfassung der Ergebnisse und definieren der Vergleichswerte\n",
    "python_MSE_data = [('GLM1', mse_glm1 , t_glm1 , 'sec' ) , # Modell, MSE, Time, Zeiteinheit\n",
    "         ('GLM2', mse_glm2 , t_glm2 , 'sec' ) ,\n",
    "         ('GLM3', mse_glm3 , t_glm3 , 'sec' ) ,\n",
    "         ('CART', mse_cart , t_cart , 'sec' ) ,\n",
    "         ('BAG', mse_bag , t_bag , 'sec' ) ,\n",
    "         ('RF', mse_rf , t_rf , 'sec' ) ,\n",
    "         ('XGB', mse_xgb , t_xgb , 'sec' ) , \n",
    "         ('BART', mse_bart , t_bart , 'sec' ) ,\n",
    "         ('MARS', mse_mars , t_mars , 'sec' )]\n",
    "\n",
    "python_MSE = pd.DataFrame(python_MSE_data, columns=[\"Methode\", \"MSE\", \"Zeit\", \"Einheit\"])\n",
    "\n",
    "r_MSE_data = [ ('GLM1', r_mse_glm1 , r_t_glm1 , 'sec'),\n",
    "              ('GLM2', r_mse_glm2 , r_t_glm2 , 'sec'),\n",
    "              ('GLM3', r_mse_glm3 , r_t_glm3 , 'sec'),\n",
    "              ('CART', r_mse_cart , r_t_cart , 'sec'), \n",
    "              ('BAG', r_mse_bag , r_t_bag , 'sec'),\n",
    "              ('RF', r_mse_rf , r_t_rf , 'sec'),\n",
    "              ('XGB', r_mse_xgb , r_t_xgb , 'sec'),\n",
    "              ('BART', r_mse_bart , r_t_bart , 'sec'),\n",
    "              ('MARS', r_mse_mars , r_t_mars , 'sec')]\n",
    "\n",
    "r_MSE = pd.DataFrame(r_MSE_data, columns=[\"Methode\", \"MSE\", \"Zeit\", \"Einheit\"])\n",
    "\n",
    "pprint(\"Die Ergebnisse in Python:\")\n",
    "pprint(python_MSE)\n",
    "pprint(\"Die Ergebnisse in R:\")\n",
    "pprint(r_MSE)"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.037258,
     "end_time": "2020-08-30T09:34:10.915856",
     "exception": false,
     "start_time": "2020-08-30T09:34:10.878598",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Sowohl in den beiden Tabellen als auch bereits durch den obigen Output der jeweiligen Modelle, sehen wir, dass der Vergleich ziemlich übereinstimmende Werte liefert. Dies zeigen auch die beiden nachfolgenden Diagramme. \n",
    "\n",
    "Im folgenden Diagramm sehen wir die Ergebnisse der jeweiligen Modelle unterschieden nach den Programmiersprachen 'Python' und 'R'. Die beiden zugehörigen Kurven verlaufen ziemlich deckungsgleich. Daraus lässt sich schließen, dass die beiden Programmiersprachen im weitestgehenden Sinne gleichwertig sind und es bei den beiden Programmiersprachen keine gibt, die sich im Rahmen des durchschnittlichen quadratischen Fehlers (MSE) besonders hervorhebt. \n",
    "\n",
    "Allerdings können wir an den beiden Tabellen ablesen, dass die Sprache 'Python' deutlich länger benötigt als die Sprache 'R', um ein entsprechendes Modell zu berechnen. Diese Unterschiede müssen aber nicht auf den beiden Programmiersprachen beruhen, da alle Berechnungen in Kaggle stattgefunden haben und wir während unserer Arbeit an diesem Projekt feststellen konnten, dass die Laufzeit dieses Jupyter-Notebooks starken Schwankungen unterliegen. Wir erklären uns dies aufgrund der unterschiedlichen Auslastungen dieser Plattform. Für uns hat es sich herausgestellt, dass eine komplette Berechnung besonders zügig durchgeführt werden konnte, wenn man diese in den frühen Morgenstunden Mitteleuopäischer Zeit angestoßen hat."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "_kg_hide-input": true,
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:11.014941Z",
     "iopub.status.busy": "2020-08-30T09:34:11.008257Z",
     "iopub.status.idle": "2020-08-30T09:34:11.198680Z",
     "shell.execute_reply": "2020-08-30T09:34:11.197911Z"
    },
    "papermill": {
     "duration": 0.245125,
     "end_time": "2020-08-30T09:34:11.198807",
     "exception": false,
     "start_time": "2020-08-30T09:34:10.953682",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot 1\n",
    "x = r_MSE['Methode']\n",
    "y1 = pd.to_numeric(r_MSE['MSE'])\n",
    "y2 = pd.to_numeric(python_MSE['MSE'])\n",
    "\n",
    "diagramm = plt.figure(figsize=(10,6))\n",
    "plt.style.use('ggplot')\n",
    "\n",
    "plt.plot(x,y1, label=\"R\",  linestyle='-', marker='o')\n",
    "plt.plot(x,y2, label=\"Python\",  linestyle='-', marker='o')\n",
    "plt.title('Übersicht der MSE')\n",
    "plt.ylabel('MSE')\n",
    "plt.xlabel('Verfahren')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.037841,
     "end_time": "2020-08-30T09:34:11.275090",
     "exception": false,
     "start_time": "2020-08-30T09:34:11.237249",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Das zweite Diagramm basiert im wesentlichen auf der Idee einer direkten Gegenüberstellung der durchschnittlichen quadratischen Fehler der beiden Programmiersprachen. Hier ist besonders hervorzuheben, dass einige Modelle durchaus bessere Werte als andere liefern.\n",
    "\n",
    "Betrachten wir zum Beispiel das Modell GLM 2 so erkennen wir, dass dieses im Vergleich zu den anderen Modellen schlechter abschneidet. Eines der Modelle, dass einen kleineren durchschnittlichen quadratischen Fehler liefert, ist zum Beispiel der Random Forest, mit dem wir uns im nächsten Abschnitt noch eingehender beschäftigen wollen."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "_kg_hide-input": true,
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:11.591979Z",
     "iopub.status.busy": "2020-08-30T09:34:11.589602Z",
     "iopub.status.idle": "2020-08-30T09:34:11.799452Z",
     "shell.execute_reply": "2020-08-30T09:34:11.798812Z"
    },
    "papermill": {
     "duration": 0.486352,
     "end_time": "2020-08-30T09:34:11.799575",
     "exception": false,
     "start_time": "2020-08-30T09:34:11.313223",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x432 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot 2\n",
    "x = r_MSE['Methode']\n",
    "y1 = pd.to_numeric(r_MSE['MSE'])\n",
    "y2 = pd.to_numeric(python_MSE['MSE'])\n",
    "\n",
    "diagramm = plt.figure(figsize=(6,6))\n",
    "plt.style.use('ggplot')\n",
    "\n",
    "plt.plot(y1,y2, 'o')\n",
    "z = np.linspace(0.012, 0.015, 1000)\n",
    "plt.plot(z, z, '-', label=\"normal\")\n",
    "plt.axis([0.012, 0.015, 0.012, 0.015])\n",
    "total_rows = y1.size\n",
    "for i in range(0, total_rows):\n",
    "    if (not (np.isnan(y1[i]) or np.isnan(y2[i]))) and y1[i] <= 0.015 and y2[i] <= 0.015:\n",
    "        plt.text(y1[i] + 0.00005 ,y2[i],str(x[i]))\n",
    "plt.title('Übersicht der MSE')\n",
    "plt.ylabel('Python')\n",
    "plt.xlabel('R')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.038141,
     "end_time": "2020-08-30T09:34:11.876309",
     "exception": false,
     "start_time": "2020-08-30T09:34:11.838168",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Abschließend lässt sich festhalten, dass wir nur kleinere Unterschiede in den Ergebnissen der einzelnen Sprachen erkennen konnten. Daraus leiten wir ab, dass keine der Sprachen bessere Ergebisse liefert als die andere und somit beide Sprachen für die Arbeit im Rahmen der aktuariellen Welt geeignet sind.\n",
    "\n",
    "Es lässt sich jedoch feststellen, dass die Modelle durchaus ihre Vorteile innerhalb verschiedener Einsatzgebiete haben. So sehen wir, dass der durchschnittliche quadratische Fehler bei einem GLM oberhalb anderer Modelle liegt, aber dieses Modell recht schnell umgesetzt werden kann und auch innerhalb einer kurzen Zeit annähernd gute Ergebnisse liefern kann. Benötigt man einen schnellen Überblick, so bietet sich die Bearbeitung mit einem GLM an.\n",
    "\n",
    "Auf der anderen Seite haben wir aufwendigere Modelle, wie das BART-Modell oder den Random Forest, trainiert. Diese sind zwar ähnlich leicht und schnell zu implementieren, haben aber deutlich längerer Laufzeiten als zum Beispiel das zuvor betrachtete GLM-Modell. Für einen zügigen Einsatz eignen sie sich daher nur bedingt. Zudem ist hier zu unterstreichen, dass beide Modelle eine Vielzahl an Parametern bieten, die die Kalibrierung des Modells erschweren. Ist man auf der Suche nach exakteren Ergebnissen, so zeigen die obigen Resultate deutlich, dass man bei diesen Modellen die richtige Wahl trifft. Allerdings muss man dort auch untersuchen, ob mit einer anderen Wahl von Parametern eine bessere Vorhersage getroffen werden kann.\n",
    "\n",
    "Im nächsten Abschnitt wollen wir verdeutlichen, wie man Vorgehen kann, wenn man sich nicht sicher ist, was die beste Einstellung dieser Parameter ist."
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.038374,
     "end_time": "2020-08-30T09:34:11.952975",
     "exception": false,
     "start_time": "2020-08-30T09:34:11.914601",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "<a id=\"AnalyseRF\"></a>\n",
    "# Analyse des Random Forest\n",
    "Die obigen Ergebnisse zeigen, dass das Modell basierend auf dem Random Forest keine wesentlich besseren MSE aufweist als zum Beispiel die GLM´s, die zu Beginn des vorherigen Abschnitts eingeführt wurden. \n",
    "Dies ist überraschend, da die Random Forrest Modelle weit verbreitet sind. Das liegt zu einem daran, dass der Lernalgorithmus relativ einfach zu bedienen ist, und zum anderen an den relativ kurzen Trainingszeiten. (vgl. obige Ergebnisse unter Berücksichtigung der gemessenen Zeiten). Ist dies der ausschlagende Grund? Oder können wir am obigen Modell noch wesentliche Änderungen vornehmen, so dass der MSE im Vergleich zu den anderen Methoden deutlich sinkt? \n",
    "Wir versuchen dies an dieser Stelle mit Hilfe von zwei Möglichkeiten dies zu erreichen. Auf der einen Seite sind die ins Modell gegebenen Trainingsdaten entscheidend für den Erfolg des maschinellenen Lernens. D.h. wir werden zunächst versuchen die Daten zu analysieren. Die Ergebnisse der Analyse wollen wir benutzen, um ein besseres Ergebnis zu erreichen. Auf der anderen Seite ist die Definition der Variablen innerhalb des Modells wichtig. Welche Variablen sind wichtig für eine Verbesserung der Ergebnisse und welche Parameter sollten eingestellt werden."
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.038196,
     "end_time": "2020-08-30T09:34:12.030382",
     "exception": false,
     "start_time": "2020-08-30T09:34:11.992186",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "<a id=\"AnalyseRF_Daten\"></a>\n",
    "## Analyse der Daten\n",
    "Wir beginnen die Analyse der Daten zunächst auf der Basis des obigen Modells. Dies soll als Ausgangspunkt für die weitere Betrachtung dienen. Zusätzlich definieren wir bereits einige weitere Größen, die wir im Laufe der Betrachtung benötigen werden."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:12.118235Z",
     "iopub.status.busy": "2020-08-30T09:34:12.117252Z",
     "iopub.status.idle": "2020-08-30T09:34:12.124772Z",
     "shell.execute_reply": "2020-08-30T09:34:12.125428Z"
    },
    "papermill": {
     "duration": 0.056643,
     "end_time": "2020-08-30T09:34:12.125585",
     "exception": false,
     "start_time": "2020-08-30T09:34:12.068942",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "RF: Erfolgreich ausgeführt\n",
      "RF: Ergebnis mse = 0.012137927171212553\n"
     ]
    }
   ],
   "source": [
    "# RF in der Analyse\n",
    "if mitBerechnung:\n",
    "    regressor = RandomForestRegressor(n_estimators=250, min_samples_split = 150, max_features=3)\n",
    "    model = regressor.fit(x_train_ohe.values, y_train.values.ravel())\n",
    "    \n",
    "    importances = model.feature_importances_\n",
    "    indices = np.argsort(importances)[::-1]\n",
    "    columns = x_train_ohe.columns\n",
    "    names = [columns[i] for i in indices]\n",
    "    \n",
    "    predictions = regressor.predict(x_test_ohe.values)\n",
    "\n",
    "    mse_rf = ((predictions - datatest['Actual_Recovery_Rate'])**2).mean(axis=None)\n",
    "else:\n",
    "    mse_rf = 0.012137927171212553\n",
    "# Output\n",
    "output(\"RF\", mse_rf, None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:12.220193Z",
     "iopub.status.busy": "2020-08-30T09:34:12.217929Z",
     "iopub.status.idle": "2020-08-30T09:34:12.895094Z",
     "shell.execute_reply": "2020-08-30T09:34:12.895670Z"
    },
    "papermill": {
     "duration": 0.730747,
     "end_time": "2020-08-30T09:34:12.895821",
     "exception": false,
     "start_time": "2020-08-30T09:34:12.165074",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1296x432 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "diagramm = plt.figure(figsize=(18,6))\n",
    "plt.style.use('ggplot')\n",
    "\n",
    "if mitBerechnung:\n",
    "    x = x_train_ohe.values.shape[1]\n",
    "    y = np.round(importances[indices], 5) \n",
    "else:\n",
    "    x = 33\n",
    "    y = [0.4701, 0.11628, 0.0861, 0.05273, 0.0446, 0.02197, 0.0158, 0.01365, 0.01278, 0.01234, 0.01036, 0.01009, 0.01004, 0.00991, 0.00803, 0.0077, 0.00769, 0.00735, 0.00732, 0.00707, 0.00657, 0.00608, 0.00569, 0.00542, 0.00529, 0.00525, 0.00524, 0.00523, 0.00514, 0.00499, 0.00488, 0.00421, 0.00414]\n",
    "    names = [\"Duration_12_4\", \"AgeBand\", \"Disability_Category_Maternity\", \"OwnOccToAnyTransition_MOD_OwnOther\", \"OwnOccToAnyTransition_MOD_Own+1\", \"Disability_Category_Injury other than back\", \"Disability_Category_Other Musculoskeletal\", \"OwnOccToAnyTransition_MOD_Own+0\", \"Gross_Indexed_Benefit_Amount_1: $ < 1000\", \"Disability_Category_Back\", \"Disability_Category_Mental and Nervous\", \"Gross_Indexed_Benefit_Amount_2: $1,000 - $1,999\", \"Disability_Category_Cancer\", \"Disability_Category_Nervous System\", \"Integration_with_STD_U: Unknown\", \"Gender_1: Male\", \"Gender_2:Female\", \"Integration_with_STD_I: Integrated with ASO of Fully-Insured STD\", \"Disability_Category_Digestive\", \"Disability_Category_Other\", \"Gross_Indexed_Benefit_Amount_6: $5,000 - $9,999\",\"Integration_with_STD_N: Not Integrated with STD\", \"Gross_Indexed_Benefit_Amount_5: $4,000 - $4,999\", \"Disability_Category_Respiratory\", \"Gross_Indexed_Benefit_Amount_9:$20,000 And Over\", \"Gross_Indexed_Benefit_Amount_7:$10,000 - $14,999\", \"Gross_Indexed_Benefit_Amount_Unknown\", \"Disability_Category_Circulatory\", \"Disability_Category_Diabetes\", \"Gross_Indexed_Benefit_Amount_4: $3,000 - $3,999\", \"Gross_Indexed_Benefit_Amount_3: $2,000 - $2,999\", \"Disability_Category_Ill-defined and Misc Conditions\", \"Gross_Indexed_Benefit_Amount_8:$15,000 - $19,999\"]\n",
    "\n",
    "\n",
    "barlist = plt.bar(range(x), y)\n",
    "\n",
    "plt.xticks(range(x), names, rotation=90, fontsize = 8)\n",
    "\n",
    "for bar in barlist:\n",
    "    yval = bar.get_height()\n",
    "    plt.text(bar.get_x() + 0.3, yval + 0.015, yval, rotation = 90)\n",
    "    \n",
    "plt.title(\"Random Forest - Importance\")\n",
    "plt.ylabel('')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.040071,
     "end_time": "2020-08-30T09:34:12.975874",
     "exception": false,
     "start_time": "2020-08-30T09:34:12.935803",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Wir sehen im vorangegangenen Diagramm die einzelnen Daten die in das Random Forest Modell eingeflossen sind. Die Analyse der Wichtigkeit der einzelnen Daten zeigt ganz deutlich, dass die Auswertung des Modells und dem daraus resultierenden durchschnittlichen quadratischen Fehler, nicht gleichmäßig auf den Daten verteilt ist, sondern, dass einige wenige Daten besonders stark in die Vorhersage eingegangen sind. Auf dieser Grundlagen wollen wir uns zunächst noch einmal dem Aufbau des Datensatzes 'recovery' nähern.\n",
    "\n",
    "Schauen wir zunächst erneut auf den verwendeten Datensatz 'recovery'. Wir stellen fest, dass nicht alle gegebenen Informationen in den Modellen aus dem vorherigen Abschnitten benutzt wurden. Durch die ausfürhliche Vorstellung der Daten sehen wir, dass folgende Spalten des Datensatzes keine Verwendung gefunden haben:\n",
    "\n",
    "* Taxability_Benefits\n",
    "* Exposures\n",
    "* Actual_Recoveries\n",
    "* Expected_Recoveries\n",
    "* Actual_Deaths\n",
    "* Expected_Deaths\n",
    "\n",
    "Wir wissen, dass zum Vorhersagen der 'Actual_Recovery_Rate' nur die weitere Information 'Taxability_Benefits' genutzt werden kann. An dieser Stelle ist nun die Frage: Wenn wir das Modell um 'Taxability_Benefits' erweitern und damit eine Vorhersage für die 'Actual_Recovery_Rate' durchführen, erhalten wir dadurch ein besseres Ergebnis?\n",
    "Dieser Fragestellung wollen wir nun nachgehen, indem wir zunächst ein Random Forest Modell ausführen, dass zusätzlich über diese Information verfügt, aber ansonsten mit dem Random Forrest Modell vergleichbar ist, dass wir zu Beginn dieses Abschnitts betrachtet haben."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:13.069895Z",
     "iopub.status.busy": "2020-08-30T09:34:13.064677Z",
     "iopub.status.idle": "2020-08-30T09:34:13.078058Z",
     "shell.execute_reply": "2020-08-30T09:34:13.077400Z"
    },
    "papermill": {
     "duration": 0.061994,
     "end_time": "2020-08-30T09:34:13.078182",
     "exception": false,
     "start_time": "2020-08-30T09:34:13.016188",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "RF_ext: Erfolgreich ausgeführt\n",
      "RF_ext: Ergebnis mse = 0.01217558698151907\n"
     ]
    }
   ],
   "source": [
    "# RF in der Analyse inkl. zusätzlicher Information\n",
    "if mitBerechnung:\n",
    "    datatrain_ext = datatrain_full[['Actual_Recovery_Rate','AgeBand', 'Duration_12_49', 'Gross_Indexed_Benefit_Amount', 'Disability_Category', 'OwnOccToAnyTransition_MOD', 'Integration_with_STD', 'Gender', 'Taxability_Benefits']]\n",
    "    datatest_ext = datatest_full[['Actual_Recovery_Rate','AgeBand', 'Duration_12_49', 'Gross_Indexed_Benefit_Amount', 'Disability_Category', 'OwnOccToAnyTransition_MOD', 'Integration_with_STD', 'Gender', 'Taxability_Benefits']]\n",
    "    \n",
    "    x_train_ext = datatrain_ext.loc[:, datatrain_ext.columns!='Actual_Recovery_Rate']\n",
    "    x_train_ext_ohe = pd.get_dummies(x_train_ext)    \n",
    "    x_test_ext = datatest_ext.loc[:, datatest_ext.columns!='Actual_Recovery_Rate']\n",
    "    x_test_ext_ohe = pd.get_dummies(x_test_ext)\n",
    "    \n",
    "    regressor_ext = RandomForestRegressor(n_estimators=250, min_samples_split = 150, max_features=3)\n",
    "    model_ext = regressor_ext.fit(x_train_ext_ohe.values, y_train.values.ravel())\n",
    "    \n",
    "    importances_ext = model_ext.feature_importances_\n",
    "    indices_ext = np.argsort(importances_ext)[::-1]\n",
    "    columns_ext = x_train_ext_ohe.columns\n",
    "    names_ext = [columns_ext[i] for i in indices_ext]\n",
    "    \n",
    "    predictions_ext = regressor_ext.predict(x_test_ext_ohe.values)\n",
    "\n",
    "    mse_rf_ext = ((predictions_ext - datatest['Actual_Recovery_Rate'])**2).mean(axis=None)\n",
    "else:\n",
    "    mse_rf_ext = 0.01217558698151907\n",
    "    \n",
    "# Output\n",
    "output(\"RF_ext\", mse_rf_ext, None)"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.040151,
     "end_time": "2020-08-30T09:34:13.158526",
     "exception": false,
     "start_time": "2020-08-30T09:34:13.118375",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Wenn wir nun die beiden Ergebnisse bzgl. der gemessenen MSE vergleichen, sehen wir, dass die Hinzunahme der weiteren Spalte aus den Daten keine wesentliche Verbesserung der Daten gebracht hat. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:13.245297Z",
     "iopub.status.busy": "2020-08-30T09:34:13.244245Z",
     "iopub.status.idle": "2020-08-30T09:34:13.250520Z",
     "shell.execute_reply": "2020-08-30T09:34:13.249744Z"
    },
    "papermill": {
     "duration": 0.051512,
     "end_time": "2020-08-30T09:34:13.250651",
     "exception": false,
     "start_time": "2020-08-30T09:34:13.199139",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-3.765981030651698e-05\n"
     ]
    }
   ],
   "source": [
    "diff = mse_rf - mse_rf_ext\n",
    "print(diff)"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.04042,
     "end_time": "2020-08-30T09:34:13.331865",
     "exception": false,
     "start_time": "2020-08-30T09:34:13.291445",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Diese Erkenntnis wird durch die Analyse des Einflusses der unterschiedlichen Parameter auf das Modell unterstrichen. Wir sehen aufgrund des \"One Hot Encoding\"-Ansatzes, hier wird die zusätzliche Information 'Taxability' mit den jweiligen Ausprägungen binär codiert, d.h. die insgesamt vier verschiedenen Ausprägungen werden in separate Informationen aufgeteilt. In nachfolgendem Diagramm sind diese Informationen grün hervorgehoben. Wir erkennen auch, dass der Einfluss dieser codierten Informationen nur marginal ist und somit auch erkennbar ist, warum der MSE der beiden Modelle etwa gleichwertig ist."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:13.437792Z",
     "iopub.status.busy": "2020-08-30T09:34:13.436892Z",
     "iopub.status.idle": "2020-08-30T09:34:14.153634Z",
     "shell.execute_reply": "2020-08-30T09:34:14.154210Z"
    },
    "papermill": {
     "duration": 0.781847,
     "end_time": "2020-08-30T09:34:14.154361",
     "exception": false,
     "start_time": "2020-08-30T09:34:13.372514",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABBEAAAKCCAYAAACH2UG8AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+j8jraAAAgAElEQVR4nOzdaXRUVdr28asyQQKExATCKGLAyIwJEeFRMBDD3AyKgIogj43I+AAqk9igMqlAg2IrShO6AUVRaaBBMBAGI8iMDNoaEVrGTIQpkBDqvB9Y1EsZMEfqVCUp/r+1WKvq1Ml9710JhFzZZx+bYRiGAAAAAAAACuFT1AMAAAAAAAAlAyECAAAAAAAwhRABAAAAAACYQogAAAAAAABMIUQAAAAAAACmECIAAAAAAABTCBEAAAAAAIAphAgAAAAAAMAUQgQAAAAAAGAKIYKbtWrVqqiHAAAAAACAJWyGYRhFPQhv0bBhQ6fnhmHoxx9/VFRUlCTpu+++K4phAQAAAABgCb+iHoA3ueuuuxQcHKyXX35ZgYGBMgxDDz30kFasWFHUQwMAAAAAwGVczmCh5cuX69FHH1X//v21d+9e3XXXXfL391eNGjVUo0aNoh4eAAAAAAAu4XIGN7hw4YLGjx+v1NRU7dq1S0ePHi3qIQEAAAAA4DJCBDfau3evtmzZogEDBhT1UAAAAAAAcBl7IrhRZGSkrly5ouzsbIWEhBT1cAAAAAAAcAl7Ilho4MCBjsdff/216tatq5EjR6pBgwZatWpVEY4MAAAAAADXcTmDhaKjo7Vr1y5JUlxcnKZPn67o6GgdOnRIjz/+uHbs2FHEIwQAAAAA4NaxEsFNzp49q+joaEnS3XffrStXrhTxiAAAAAAAcA17Iljohx9+UMOGDWUYhg4fPqzTp08rNDRUdrtdly9fLurhAQAAAADgEkIEC33//fdOz8uWLStJysrK0quvvloUQwIAAAAAwDLsiQAAAAAAAExhTwQLnT17VmPGjFHv3r21ePFip9euv3MDAAAAAAAlESGChZ555hkZhqFHH31UH3/8sR599FHl5uZKkrZu3VrEowMAAAAAwDWECBb6+eefNXXqVHXp0kXLly9XdHS0WrVqpczMzKIeGgDAYjabTQsXLizqYQAAAHgUIYKFcnNzZbfbHc/HjRun/v37q0WLFgQJAGChvn37ymazyWazydfXV9WqVdPTTz+tY8eOFfXQPOLhhx92zP/6P0W96u3111/XXXfdZVm9DRs2yGaz6ejRo5bVdIf4+Hj17du3qIcBAIBHECJYqFOnTlq/fr3TsT59+mj69OkKCAgoolEBgHd66KGHdOLECf33v//V4sWLtXv3bnXv3r2oh+UxTzzxhE6cOOH0JyYm5pZq2e12XblyxeIRer/Lly+L/akBALcbQgQLvfHGG4qPjy9wvG3btvrpp5+KYEQA4L0CAgJUqVIlVa1aVS1atFD//v21ZcsWnT171nHOuHHjVKdOHQUFBal69eoaMGCAzpw543g9MTFRfn5+SklJUXR0tIKCghQbG6udO3c69UpOTlbDhg1VunRpNWzYUMnJyQXG85///EcdOnRQ2bJlVbZsWXXq1EmpqakFeiUnJ6tBgwYKDAxUy5Ytdfz4cW3atEn33XefypQpo/j4eFMrKgIDA1WpUiWnP/7+/pKu7sPTokULBQYGKjQ0VE888YTS0tIcHzthwgTVqlVLS5Ys0b333quAgAB9//33On/+vIYNG6aqVasqKChI9913nz7//HOnvpMnT9bdd9+tUqVKqUKFCmrTpo0uXryoxMREjR8/XkeOHHGsjJgwYUKh8/gjrq1MWLVqlZo1a6bAwEDFxMTowIEDOnDggB588EEFBQXp/vvv18GDBx0fd+29T0pKUr169VS6dGndf//92rVrl1P9VatWKSYmRqVKlVLFihU1cOBAXbhwwfF63759FR8fr7ffflt33XWXSpUqpe7du2vdunVasGCBY94bNmyQZN3X388//6zu3bvrjjvuUFBQkBo2bKiVK1c6Xt+5c6cSEhJUtmxZVahQQd26ddORI0esfOsBAHAgRLDYtm3btH37dknSwYMHNWPGDK1ataqIRwUA3u348eNaunSpfH195evr6zgeGBiouXPn6uDBg0pMTNSGDRs0dOhQp4+12+0aM2aMZs2apV27dik0NFSPP/648vPzHbU7duyomJgY7dq1S9OnT9ewYcOcaly8eFEJCQm6dOmSNm7cqI0bN+r8+fNq27at8vLynHpNnDhRH374oVJSUnT8+HH16NFDr7zyiv72t7/p66+/1tGjRzVixIhbfi9OnjyphIQEVatWTdu2bdOKFSu0f/9+PfroowXes3fffVeJiYk6ePCgatSooU6dOmnv3r1asmSJ9u/fr+eff149e/bUunXrJEmff/65pk6dqlmzZumnn37SV199pXbt2kmSevTooVGjRqlatWqOlREvvPDCLc/j94wbN06TJk3Szp07FRAQoF69eun555/XxIkTHceeeeYZp4+x2+166aWX9O6772rbtm2qWLGiOnTooJycHEnSd999pz/96U9q0aKF9uzZowULFmjlypUaMGCAU51t27Zp/fr1WrZsmfbu3at58+bpoYce0uOPP+6Yd/PmzSVZ8/V38uRJNW/eXKdPn9by5cu1b98+vfbaa/LxufpfuIMHD6ply5Zq1qyZduzYofXr18vX11ePPPKILl265Jb3HwBwmzNgmQkTJhhNmzY1YmJijNGjRxtxcXHGxIkTjYceesh4/fXXi3p4AOA1+vTpY/j6+hplypQxAgMDDUmGJGPkyJG/+3Gff/65ERAQYFy5csUwDMOYP3++IcnYuXOn45wtW7YYkowffvjBMAzDGDdunHHnnXcaly9fdpyzYsUKQ5Lxz3/+0zAMw/jwww+NwMBAIz093XHOyZMnjdKlSxsLFixw6rV7927HOW+88YYhydixY4fj2IwZM4ywsLDfnUfLli0NPz8/o0yZMo4/cXFxhmEYxssvv2xUrVrVyM3NdZy/Z88eQ5KxceNGwzAM4y9/+Yths9mMI0eOOM5JTk42SpUqZWRnZzv1euaZZ4zOnTs7xla7dm0jLy/vhuN67bXXjBo1avzu2P+I5ORkQ5Lx66+/Oj3/4osvHOd88sknhiRj6dKljmOff/65Ick4d+6cYRj//71PSkpynJOVlWWUKVPG+OCDDwzDMIynnnrKiI2Ndeq/bNkyw2azGYcPHzYM4+rXXfny5R11r2ndurXRp0+fQudzK19/L7/8shEREWGcP3/+hjX79Olj9OjRw+nYpUuXjMDAQKf3CQAAq/gVQW7htZYuXao9e/YoNzdXlSpV0tGjRxUcHKwXX3xRTZs21bhx44p6iADgNZo2baoFCxbo0qVL+uSTT/TVV1/ptddeczrn888/11//+lelpqbq7NmzstvtysvL08mTJ1WlShVJV++y0KhRI8fHVK1aVZJ06tQpRUVF6eDBg7r//vvl5/f/v2U++OCDTn0OHDigunXrKjw83HEsIiJCUVFROnDggOOYzWZTgwYNHM8rVaokSWrYsKHTsczMTF25csVpVcVvde3aVZMnT3Y8L126tGMsDzzwgNNePI0aNVL58uV14MABtWjRwjG+O++803HO9u3blZeX55j/NXl5eapdu7Yk6fHHH9fs2bNVo0YNJSQkqHXr1urSpYvKlSt303HeSLt27bR582bH8/Pnz/+hj7/+83Wz91CS0tLSVLZsWcfxZs2aOR6HhoaqTp06jsseDhw4oFatWjn1admypQzDcKzUkKQ6deo41fw9Vnz97dy5U82bN1eZMmVu2GP79u1KTU0tMKZLly5xKSUAwC0IESzk5+cnX19fBQUFKTIyUsHBwZKuLme8tuwQAGCNwMBA1apVS5JUv359/fjjjxo0aJD+/ve/S5K+/fZbde/eXWPGjNGbb76p0NBQbd26VX369HG6xMDHx8fph3WbzSZJjrvtGIbhOPbbcwo79tuPvVmva3sZXH/MKGTDvuDgYMf8zYzlt8d/+0Op3W5X+fLlHZfkXe9aIFG1alX98MMPSk5O1vr16/Xaa69p1KhR+vbbb1W9evXfHe/1PvzwQ128eNH0+b91o/frRseuv2PSjfz2Pb6V9+1mrPr6+71xXTuvd+/eGj16dIHXwsLCTI0VAIA/gp9sLRQQEOC4tvL6TZHOnDlDiAAAbjZhwgQtWLBAO3bskCR9/fXXCg8P1+uvv66mTZvqnnvuuaVbBdarV0/ffvut090Lvv766wLnHDhwQBkZGY5jp06d0o8//qh69erd4oxuTb169bRlyxanH1T37t2rM2fO/O5YmjRpouzsbF26dEm1atVy+nP9ioVSpUqpbdu2euONN7Rv3z7l5ORo2bJlkq5+HzRzl4eqVas61feU62+BmZ2drR9++EF16tSRdPV927hxo9P5GzdulM1mU926dX+37o3mbdXXX0xMjFJSUpw2eLxekyZN9N133ykyMrLA5y00NPQP9wMAoDD8ZGuhTZs2KSgoSJKcQoPLly9rwYIFRTUsALgt3HvvverYsaPGjBkjSYqKilJ6errmzZunQ4cO6R//+IfefffdP1z3+eefV3p6uvr376/vv/9e69atK3B52hNPPKEKFSqoR48e2rVrl3bu3KmePXuqatWq6tGjhyXzM2vw4ME6e/as+vbtq/379+vrr79W79699eCDD+qhhx666ce1atVK8fHx6tatm7744gsdOnRIO3fu1Ntvv60PPvhAkjRv3jx98MEH2rt3r44cOaJFixbp3Llzjh+ya9asqZMnT2rLli3KyMhwBOvFgc1m00svvaRNmzZp3759evrpp1WmTBk98cQTkqQXX3xRu3bt0ogRI/TDDz/oyy+/1JAhQ/Tkk086hSg3UrNmTe3cuVM///yzMjIydPnyZcu+/gYOHCi73a7OnTsrJSVFv/zyi1auXKnVq1dLksaOHavvv/9eTz31lLZt26ZffvlFycnJGjZsmA4dOvTH3ygAAApBiGChUqVKFTi2fPlyhYeHO10DCwBwj5deeklJSUlat26dOnbsqHHjxmns2LFq0KCBPv74Y7355pt/uGbVqlW1YsUKbdu2TY0bN9awYcM0Y8YMp3MCAwO1du1alSpVSi1atFDLli1VpkwZffnll057E3hCRESE1q5dq6NHjyo2NlYdO3ZU/fr19dlnn/3ux9lsNi1fvlzdunXTiBEjdO+996pDhw7697//rcjISElX9xGYP3++Hn74YdWpU0czZszQ3Llz1bp1a0lSly5d1L17d3Xo0EEVKlTQG2+84fb5muXj46PJkyfrueeeU5MmTXTixAn9+9//dlye0LBhQy1fvlwbN25Uo0aN1Lt3b3Xo0EHvvfdeobVHjhyp8PBwNWrUSBUqVFBKSoplX3+VK1fW119/rXLlyql9+/aqV6+exo0b57gUo06dOvrmm290/vx5tWnTRnXr1tWf//xnXbx4USEhIX+4HwAAhbEZhV10CdN+ey9twzA0aNAgx28eunXrVhTDAgDgtpaYmKhnn33WcdtEAABw69hY0UKPP/642rZtq4oVKzp+Q3DhwgWtWLFCNpuNEAEAAAAAUKIRIlhoy5YtGj16tGJjYzVgwADZbDZt2LBB8+fPL+qhAQAAAADgMvZEsFBsbKy++uor5eXlqVWrVtq2bdvv3pYJAAC4X9++fbmUAQAAi7AngpscP35c//d//6cdO3awOzIAAAAAwCsQIgAAAAAAAFO4nMFD5s6dW9RDAAAAAADAJUW6seLx48eLsr1bhYeHKyMjw/H89OnTls/3tz2s5u769ChePbxhDvQoXj28YQ70KD716VG8enjDHOhRvHp4wxzoUXzqe1OPolSlSpUbHufuDBZLTU3VmjVrlJ2drdzcXEVERCghIUG9e/cu6qEBAAAAAOASLmew0Jw5c/T888/LMAzFxsaqUaNGMgxDAwcO1DvvvFPUwwMAAAAAwCWsRLDQRx99pOTkZPn7+zstbenfv7/i4uI0ePDgIh4hAAAAAAC3jpUIFvLx8dGpU6cKHE9LS5OPD281AAAAAKBkYyWChSZOnKgePXqoZs2aqlmzpnJzc3Xs2DEdPnxYr7/+elEPDwAAAAAAlxAiWCguLk6bN2/W7t27lZOTozNnzqhy5cpq3LixfH19i3p4AAAAAAC4hBDBYj4+PoqJifH6230AAAAAAG4/XKgPAAAAAABMIUQAAAAAAACmECIAAAAAAABTCBEAAAAAAIAphAgAAAAAAMAUQgQAAAAAAGAKt3j8g678+U+mzjtlsp7vB8tvfTAAAAAAAHgQKxEAAAAAAIAphAgAAAAAAMAUQgQAAAAAAGAKIQIAAAAAADCFEAEAAAAAAJhCiAAAAAAAAEwhRAAAAAAAAKYQIgAAAAAAAFMIEQAAAAAAgCmECAAAAAAAwBRCBAAAAAAAYAohAgAAAAAAMIUQAQAAAAAAmEKIAAAAAAAATCFEAAAAAAAAphAiAAAAAAAAUwgRAAAAAACAKYQIAAAAAADAFEIEAAAAAABgiqkQYc+ePRo2bJiGDBmiZcuW3fS81NRU9ejRQ1u3brVsgAAAAAAAoHgoNESw2+2aN2+exo4dq5kzZyolJUVHjx694XmLFi1S48aN3TJQAAAAAABQtAoNEVJTU1WpUiVFRETIz89PzZs31/bt2wuct3r1ajVt2lTBwcFuGSgAAAAAAChahYYIWVlZCgsLczwPCwtTVlZWgXO2bdumhIQE60cIAAAAAACKBb/CTjAMo8Axm83m9DwxMVFPPvmkfHx+P5NISkpSUlKSJGnq1KkKDw//I2MtFk5ZXM+V98DPz8+t76G769OjePXwhjnQo3j18IY50KP41KdH8erhDXOgR/Hq4Q1zoEfxqe9NPYqjQkOEsLAwZWZmOp5nZmYqNDTU6Zyff/5Zs2bNkiSdPXtWu3fvlo+Pj+6//36n8+Lj4xUfH+94npGR4dLgvYEr70F4eLhb30N316dH8erhDXOgR/Hq4Q1zoEfxqU+P4tXDG+ZAj+LVwxvmQI/iU9+behSlKlWq3PB4oSFCZGSkTpw4obS0NN1xxx365ptvNHToUKdz5syZ4/Q4JiamQIAAAAAAAABKtkJDBF9fX/Xr10+TJk2S3W5XXFycqlevrrVr10oS+yAAAAAAAHCbKDREkKTo6GhFR0c7HbtZeDBo0CDXRwUAAAAAAIqdQu/OAAAAAAAAIBEiAAAAAAAAkwgRAAAAAACAKYQIAAAAAADAFEIEAAAAAABgCiECAAAAAAAwhRABAAAAAACYQogAAAAAAABMIUQAAAAAAACmECIAAAAAAABTCBEAAAAAAIAphAgAAAAAAMAUQgQAAAAAAGAKIQIAAAAAADCFEAEAAAAAAJhCiAAAAAAAAEwhRAAAAAAAAKYQIgAAAAAAAFMIEQAAAAAAgCmECAAAAAAAwBRCBAAAAAAAYAohAgAAAAAAMIUQAQAAAAAAmEKIAAAAAAAATCFEAAAAAAAAphAiAAAAAAAAUwgRAAAAAACAKYQIAAAAAADAFEIEAAAAAABgCiECAAAAAAAwhRABAAAAAACYQogAAAAAAABMIUQAAAAAAACmECIAAAAAAABTCBEAAAAAAIAphAgAAAAAAMAUQgQAAAAAAGAKIQIAAAAAADCFEAEAAAAAAJhCiAAAAAAAAEwhRAAAAAAAAKYQIgAAAAAAAFMIEQAAAAAAgCmECAAAAAAAwBRCBAAAAAAAYAohAgAAAAAAMIUQAQAAAAAAmEKIAAAAAAAATCFEAAAAAAAAphAiAAAAAAAAUwgRAAAAAACAKYQIAAAAAADAFEIEAAAAAABgCiECAAAAAAAwhRABAAAAAACYQogAAAAAAABMIUQAAAAAAACmECIAAAAAAABTCBEAAAAAAIAphAgAAAAAAMAUQgQAAAAAAGAKIQIAAAAAADCFEAEAAAAAAJjiZ+akPXv2aP78+bLb7WrdurW6dOni9Pr27du1ZMkS2Ww2+fr6qm/fvrr33nvdMmAAAAAAAFA0Cg0R7Ha75s2bp5dffllhYWEaM2aMmjRpomrVqjnOadCggZo0aSKbzaYjR45o5syZ+utf/+rWgQMAAAAAAM8q9HKG1NRUVapUSREREfLz81Pz5s21fft2p3NKly4tm80mScrNzXU8BgAAAAAA3qPQlQhZWVkKCwtzPA8LC9NPP/1U4Lxt27Zp8eLFOnPmjMaMGWPtKAEAAAAAQJGzGYZh/N4JW7Zs0d69ezVgwABJ0qZNm5Samqp+/frd8PyDBw/qs88+0/jx4wu8lpSUpKSkJEnS1KlTlZeX5+r4Pe5U1+aW1ov44ptb/lg/Pz/l5+dbOBrP1qdH8erhDXOgR/Hq4Q1zoEfxqU+P4tXDG+ZAj+LVwxvmQI/iU9+behSlgICAGx4vdCVCWFiYMjMzHc8zMzMVGhp60/Pr1q2rOXPm6OzZswoODnZ6LT4+XvHx8Y7nGRkZhQ7c27nyHoSHh7v1PXR3fXoUrx7eMAd6FK8e3jAHehSf+vQoXj28YQ70KF49vGEO9Cg+9b2pR1GqUqXKDY8XuidCZGSkTpw4obS0NOXn5+ubb75RkyZNnM45efKkri1oOHTokPLz81WuXDkLhg0AAAAAAIqLQlci+Pr6ql+/fpo0aZLsdrvi4uJUvXp1rV27VpKUkJCgrVu3atOmTfL19VVAQICGDx/O5ooAAAAAAHiZQkMESYqOjlZ0dLTTsYSEBMfjLl26qEuXLtaODAAAAAAAFCuFXs4AAAAAAAAgESIAAAAAAACTCBEAAAAAAIAphAgAAAAAAMAUQgQAAAAAAGAKIQIAAAAAADCFEAEAAAAAAJhCiAAAAAAAAEwhRAAAAAAAAKYQIgAAAAAAAFMIEQAAAAAAgCmECAAAAAAAwBRCBAAAAAAAYAohAgAAAAAAMIUQAQAAAAAAmEKIAAAAAAAATCFEAAAAAAAAphAiAAAAAAAAUwgRAAAAAACAKYQIAAAAAADAFEIEAAAAAABgCiECAAAAAAAwhRABAAAAAACYQogAAAAAAABMIUQAAAAAAACmECIAAAAAAABTCBEAAAAAAIAphAgAAAAAAMAUQgQAAAAAAGAKIQIAAAAAADCFEAEAAAAAAJhCiAAAAAAAAEwhRAAAAAAAAKYQIgAAAAAAAFMIEQAAAAAAgCmECAAAAAAAwBRCBAAAAAAAYAohAgAAAAAAMIUQAQAAAAAAmEKIAAAAAAAATCFEAAAAAAAAphAiAAAAAAAAUwgRAAAAAACAKYQIAAAAAADAFEIEAAAAAABgCiECAAAAAAAwhRABAAAAAACYQogAAAAAAABMIUQAAAAAAACmECIAAAAAAABTCBEAAAAAAIAphAgAAAAAAMAUQgQAAAAAAGAKIQIAAAAAADCFEAEAAAAAAJhCiAAAAAAAAEwhRAAAAAAAAKYQIgAAAAAAAFMIEQAAAAAAgCmECAAAAAAAwBRCBAAAAAAAYAohAgAAAAAAMMXPzEl79uzR/PnzZbfb1bp1a3Xp0sXp9c2bN+tf//qXJKl06dJ69tlnddddd1k+WAAAAAAAUHQKXYlgt9s1b948jR07VjNnzlRKSoqOHj3qdE7FihU1YcIEvfXWW3r00Uc1d+5ctw0YAAAAAAAUjUJDhNTUVFWqVEkRERHy8/NT8+bNtX37dqdzoqKiVLZsWUlS7dq1lZmZ6Z7RAgAAAACAIlNoiJCVlaWwsDDH87CwMGVlZd30/PXr1+u+++6zZnQAAAAAAKDYsBmGYfzeCVu2bNHevXs1YMAASdKmTZuUmpqqfv36FTh3//79mjdvnl599VWVK1euwOtJSUlKSkqSJE2dOlV5eXlWzMGjTnVtbmm9iC++ueWP9fPzU35+voWj8Wx9ehSvHt4wB3oUrx7eMAd6FJ/69ChePbxhDvQoXj28YQ70KD71valHUQoICLjh8UI3VgwLC3O6PCEzM1OhoaEFzjty5Ijef/99jRkz5oYBgiTFx8crPj7e8TwjI6PQgXs7V96D8PBwt76H7q5Pj+LVwxvmQI/i1cMb5kCP4lOfHsWrhzfMgR7Fq4c3zIEexae+N/UoSlWqVLnh8UIvZ4iMjNSJEyeUlpam/Px8ffPNN2rSpInTORkZGXrrrbc0ePDgmzYCAAAAAAAlW6ErEXx9fdWvXz9NmjRJdrtdcXFxql69utauXStJSkhI0NKlS3X+/Hl9+OGHjo+ZOnWqe0cOAAAAAAA8qtAQQZKio6MVHR3tdCwhIcHxeMCAAY49EwAAAAAAgHcq9HIGAAAAAAAAiRABAAAAAACYRIgAAAAAAABMIUQAAAAAAACmECIAAAAAAABTCBEAAAAAAIAphAgAAAAAAMAUQgQAAAAAAGAKIQIAAAAAADCFEAEAAAAAAJhCiAAAAAAAAEwhRAAAAAAAAKYQIniRmTNnFvUQAAAAAABejBDBiyxevLiohwAAAAAA8GJ+RT0A/DFRUVGOxzabTYZhSJIMw9ClS5eKalgAAAAAgNsAIUIJExwcrFWrVqlChQoKDw9XRkaG47UmTZoU4cgAAAAAAN6OyxlKmMcee0xHjx694Wtdu3b18GgAAAAAALcTViKUMKNGjbrpa+PGjfPgSAAAAAAAtxtChBJq7969On/+vC5cuKC7775btWrVKuohAQAAAAC8HCFCCbNlyxa9+uqrCg4O1v79+xUTE6MzZ87I399fs2bNUtWqVYt6iAAAAAAAL8WeCCXMX/7yFy1cuFBLlizR1q1b5e/vr3/9618aOnSoXnjhhaIeHgAAAADAixEilDB2u11hYWGSpDvvvNOxyWKLFi108uTJohwaAAAAAMDLcTlDCdOwYUONHDlSDz74oDZs2KBmzZpJki5evKgrV64U8egAAAAAAN6MEKGEmTZtmhYtWqQdO3aoVatW6tixo+O1xYsXF+HIAAAAAADejhChhPH391ffvn0lSeHh4crIyJAkBQYGqlq1akU4MgAAAACAtyNEKGHOnj2rd955R2vWrFFWVpbsdrvCw8PVpk0bDRo0SOXLly/qIQIAAAAAvBQbK5YwAwYMUPny5UPLfugAACAASURBVPXpp5/qxIkTOnDggD799FOVL19ezz33XFEPDwAAAADgxQgRSphff/1VgwYNUsWKFR3HKlasqEGDBunYsWNFODIAAAAAgLcjRChhqlWrpnfffVfp6emOY+np6ZozZ46qVKlShCMDAAAAAHg7QoQS5m9/+5tOnz6tRx99VJUqVVK9evX02GOPKTs7W++9915RDw8AAAAA4MXYWLGECQkJ0bhx4zRu3DinuzMAAAAAAOBurETwIkuWLCnqIQAAAAAAvBghghd56623LK13+fLlAseysrIs7QEAAAAAKDm4nKGEiY+Pdzz29fXVlStXJEmGYVh2aUNKSoqGDh2q/Px81a1bV2+88YaqV68uSerVq5fWrFljSR8AAAAAQMlCiFDCpKena9GiRQoJCVFoaKhOnz4t6WqI0LlzZ0t6TJo0SYsXL9b//M//KDExUT179tTs2bMVExMjwzAs6QEAAAAAKHkIEUqY+Ph45eTkqH79+goPD1eZMmUcrzVr1sySHnl5eYqKipIkdezYUbVr19azzz6rsWPHymazWdIDAAAAAFDyECKUMNOnT7/pa3PmzLGkh7+/v9LS0hQeHi5JioqK0pIlS9SnTx8dOXLEkh4AAAAAgJKHjRW9yIULFyypM2bMGKWnpzsdq1Klij777DMNGjTIkh4AAAAAgJKHEMGLPPzww5bUadGiherVq1fgeHBwsIYNG2ZJDwAAAABAycPlDCXM+++/73hcpkwZx+oDwzCUk5Pj9v5PPfWUFi5c6PY+AAAAAIDihxChhJk2bZoGDBggPz+/AsGB3W63pMe+ffskSSEhIcrOznYcNwxDBw4csKQHAAAAAKDkIUQoYerXr6+2bduqYcOGCg8PV0ZGhuO1xYsXW9Kjffv2euCBB+Tv76+8vDyn186ePWtJDwAAAABAyUOIUMLMmDFDd9xxxw1fW716tSU9ateurWnTpun+++93CikkqUmTJpb0AAAAAACUPIQIJUytWrVu+lqFChUs6TFixIibXhrx+uuvW9IDAAAAAFDycHcGL2LVhocdO3a8aVjRtm1bS3oAAAAAAEoeViJ4EcMwLKt1+PBh/eMf/1Bqaqp8fX1Vs2ZNdenSRcHBwZb1AAAAAACULIQIJVBqaqrWrFmj7Oxs5ebmKiIiQgkJCerdu7cl9efNm6evvvpKrVu31p49e1SvXj0dP35cf/rTnzR58mQ1b97ckj4AAAAAgJKFyxlKmDlz5uj555+XYRiKjY1Vo0aNZBiGBg4cqHfeeceSHosXL9aiRYs0ZswYLVmyRD/99JNGjx6tRYsWacKECZb0AAAAAACUPKxEKGE++ugjJScny9/f3+kWj/3791dcXJwGDx5sSZ/8/HxJUm5urs6fPy9Jqlq1quM4AAAAAOD2Q4hQwvj4+OjUqVOqVq2a0/G0tDT5+FizsKRXr17q0KGDmjVrpo0bN2rQoEGSpMzMTIWEhFjSAwAAAABQ8hAilDATJ05Ujx49VLNmTdWsWVO5ubk6duyYDh8+bNntF5999lk99NBDOnXqlPr06eO4U0NYWJg+//xzS3oAAAAAAEoeQoQSJi4uTps3b9bu3buVk5OjM2fOqHLlymrcuLF8fX0t6xMVFaVatWrp4MGD2r9/vyIiIlShQgXL6gMAAAAASh5ChBLIx8dHMTExTnsiWGn//v0aPXq0cnJyHMHBiRMnFBwcrClTpqhBgwYu90hJSdGqVauUkZEhu92umjVrqlevXqpZs6bLtQEAAAAA7kGIgAKGDx+uadOmKSEhwSmk2Llzp4YPH66kpCSX6k+ePFnp6el68MEHlZycrIiICNWoUUPPPfechgwZok6dOrk6BQAAAACAG3CLRxSQk5Oj6OjoAsdjYmJ08eJFl+uvW7dOM2fO1KOPPqqFCxdqx44devLJJ/XJJ59o5syZLte/me7du7utNgAAAADcDliJgAJatWql3r17q1+/fipbtqwk6fjx41q6dKkefvhhl+vbbDadPn1aoaGhOn78uK5cuSJJCgkJkWEYLteXpPj4eMdjX19f5efn69ChQ47jrq6mAAAAAIDbESECCnjttde0fv16bdiwQYcPH5ZhGKpcubL69u2r1q1bu1x/yJAhatOmjSIjI/XLL7847iqRmZmpunXrulxfkqpVq6Zy5cpp2LBhqly5srKystS1a1clJiZaUh8AAAAAbkeECLihVq1a6fHHH3fLxo2dO3dWixYt9N///lcxMTHKz8+XdPUWknPmzLGkR2JiolavXq2XXnpJL774opo1ayZ/f39Vq1bNkvoAAAAAcDtiTwT8IQsXLrSkTmhoqBo1aqSQkBCn46mpqZbUl6R27dpp4cKF2rRpk/r27au8vDzLagMAAADA7YiVCMXQlT//ydR5p0zW8/1g+a0P5jes2rPgZnr16qXt27dbVi8oKEhvvvmmNm7cqJ07d1pWFwAAAABuR4QIuKHU1FQlJibq559/ls1mU0REhBISEtS7d2+Xa48fP97xuHTp0rp06ZKkqwHF2bNnXa5/I/Xq1bM0nAAAAACA2xEhAgqYM2eOli1bpieeeEKNGzeWJJ04cUIDBw5U586dNXjwYJfqL1myRK+88ooCAgJUrlw5nTt3zvHasmXLXKp9zfvvv+94XKZMGV24cEFvv/22cnNzJUnPPfecJX0AAAAA4HZCiIACPvroIyUnJ6ty5cpOGyv2799fcXFxLocIjRo1UlRUlGJjYxUeHu7UY8aMGS7Vvmb69Olq1aqVoqKiZBiGcnJyZLfbdeHCBUvqAwAAAMDtyFSIsGfPHs2fP192u12tW7dWly5dnF4/duyY3n33Xf3yyy/q2bOn/vQnc9f0o3jy8fHRqVOnVLlyZafjaWlp8vFxfS/OuXPnqnTp0jd8bevWrS7Xl6Tk5GRNmDBBOTk5GjdunHJycvTpp59qxIgRltQHAAAAgNtRoSGC3W7XvHnz9PLLLyssLExjxoxRkyZNnG6VV7ZsWT3zzDNcc+4lJk6cqB49euiee+5RhQoVJF0Nig4fPqzXX3/d5fqhoaEFjmVkZCg8PNzl2tdUrVpVH3zwgdasWaP27durb9++ltUGAAAAgNtVoSFCamqqKlWqpIiICElS8+bNtX37dqcQoXz58ipfvrx27drlvpHCY+Li4rR582b98ssv+uGHH2QYhipXrqzGjRvL19fX5fqnT592PPbx8VFWVpY6dOigNWvWyDCMG4YMt6pNmzbq2rWrxowZU2BlBQAAAADgjyk0RMjKylJYWJjjeVhYmH766Se3DgpFz8fHR02bNlVkZKTltRs0aOAIoXx8fGS323Xy5Em1bdtWNptNW7ZssbRfUFCQ0x0hAAAAAAC3ptAQwTCMAsdsNtstNUtKSlJSUpIkaerUqZYuX/eUUxbXu9F74IkeZvj5+bnlczR58mStX79eU6dOVePGjZWfn6977rlHP/74o+W9pILz6NSpk1asWOHWHu7g7h7eMAd6FK8e3jAHehSf+vQoXj28YQ70KF49vGEO9Cg+9b2pR3FUaIgQFhamzMxMx/PMzMxbXm4eHx+v+Ph4x/Prd+W/XXniPbjVHr+9c4JVnn76acXHx+uVV15RZGSkBg0aJMMwLO21b98+x+OQkBBlZ2dLuhqK7dmzx/J5ueu98mQPb5gDPYpXD2+YAz2KT316FK8e3jAHehSvHt4wB3oUn/re1KMoValS5YbHCw0RIiMjdeLECaWlpemOO+7QN998o6FDh1o+QNxeqlSporlz52rr1q3q2bOnLl68aGn99u3b64EHHpBhGAoICFBeXp7jtbNnz1raCwAAAABuF4WGCL6+vurXr58mTZoku92uuLg4Va9eXWvXrpUkJSQkKDs7W6NHj9bFixdls9m0atUqzZgxQ0FBQW6fAEq2jh07qlGjRjpy5IildWvXrq1p06bp7rvvLpAQNmnSxNJeAAAAAHC7KDREkKTo6GhFR0c7HUtISHA8DgkJ0XvvvWftyHDb2Ldvn3bv3q1Tp06pZcuWltQcMWKE7Hb7DV+z4jaVAAAAAHA78inqAeD206FDB8fjefPmady4cbpw4YJmzJihd955x5IeHTt2VK1atW74Wtu2bS3pAQAAAAC3G0IEeNzly5cdj+fNm6ePP/5YI0aM0EcffaQvvvjCLT23bdum999/Xxs3bnRLfQAAAAC4HZi6nAHe5cqf/2TqPLO3mvT9YPkf6m+325WdnS273S7DMBQWFiZJCgoKkq+v7x+qdTMdOnTQv//9b0lXg4p33nlH7dq104wZM7Rv3z4NHjzYkj4AAAAAcDshRIDHnTt3Tu3atZNhGPL19VVaWpoqVqyoCxcuyDAMS3rcaLVDWFiYBgwYoE6dOhEiAAAAAMAtIESAx3377beOx9ffOcHHx0fz5s2zpIcnVjucPn1aiYmJioiI0JAhQzRr1izt3LlTtWvX1pAhQxQSEmJJHwAAAAAoLtgTAUUmPT1du3fv1v79+5Wenq7AwEDdeeedltS+ttqhffv2ysrKUlpamiRZutph6NChysnJ0XfffaeEhASlp6dr0KBBKl26tIYPH25JDwAAAAAoTliJAI/bv3+/Ro8erXPnzql69eq6fPmyTpw4oeDgYE2ZMkUNGjRwuYcnVjucPHlS//znP2UYhu6//3599NFHkqSmTZvqkUcesaQHAAAAABQnhAjwuOHDh2vatGmKjo52+gF/586dGj58uJKSkizrlZ6erl9//VVnzpxRRESEKlSoYNlqB8MwlJ2drQsXLuj8+fP69ddfVb16dWVlZTntyQAAAAAA3oIQAR6Xk5Oj6OjoAsdjYmJ08eJFS3p4YrXD4MGD1bJlS0nS3Llz9cILL8hms+mnn36y7HIG9l0AAAAAUJwQIsDjWrVqpd69e+uxxx5T3bp1lZ2drePHj2vp0qV6+OGHLenhidUOXbp0UadOnWQYhipVqqRmzZrpwIEDqlSpkiIiIlyuL13dd+Hee+917LsQGRmpQYMGadOmTRo+fLjmz5/vco/Vq1frgQceUHh4uDIzMzVx4kQdOHBAtWvX1iuvvKIqVapYMBMAAAAA3oAQAR732muvaf369VqzZo1Wrlyp3NxcVa5cWX379lXr1q0t6eGJ1Q6SnO70kJubKx8fH5UuXdqy+p7Yd2HatGnasGGDJGncuHGKjo7W6NGjtXnzZo0YMUIff/yxyz1SUlK0atUqZWRkyG63q2bNmurVq5dq1qzpcm0AAAAAnkOIgCLRqlUrtWrVymmVgNX13b3aYcyYMZoyZYqkqz8kP/XUU6pRo4YOHz6sadOmWRKIeGLfhStXrjgeHz58WO+9954kqUePHvrwww9drj958mSlp6frwQcfVHJysiIiIlSjRg0999xzGjJkiDp16uRyD0+YMGGCOnTooNjY2KIeCgAAAFBkCBFQrCxcuFBPPfWUy3U8sdph165djscTJ07U3//+dzVo0EBHjhzRgAEDLOnjiX0XmjVrpjfffFMTJkxQs2bN9OWXX6pt27ZKSUlRcHCwy/XXrVundevWSZL+93//Vy1bttT48ePVoUMHdevWrcSECJ999pm+/fZbZWZmqkePHmrTpo3q169vaQ9PBBXu3mcjKytLd9xxh+P5Z599pj179igqKkpPPvmkbDabq1PwSA8AAADcGCECihXDMCyr5e7VDtc7e/asY7PGGjVqOP123xWe2Hdh0qRJmj17tho0aKArV67ogw8+UFBQkB555BHNnj3b5fo2m02nT59WaGiojh8/7nhvQkJCLP18u1vlypW1evVqHTp0SElJSRoyZIiuXLmiLl26qHPnzoqMjHS5hyeCCnfvs9GrVy+tWbNGkjRlyhQlJyerS5cuSkpK0k8//aSJEye6PAdP9AAAAMCNESKgSKSmpmrNmjXKzs5Wbm6uIiIilJCQoN69e7u9t1WrHVJTUxUfHy/DMHT06FFlZ2crJCREdrtd+fn5Foz0Kl9fX8elC35+fmrUqJGkgr+NvVX+/v4aOXKkpkyZokOHDik/P9+SutcMGTJEbdq0UWRkpH755Re9/vrrkqTMzEzVrVvXsj7udu2323fffbfGjh2r/v376+DBg/rXv/6lp59+WikpKS738ERQ4e59Nq4PhpYtW6ZPP/1UQUFB6tq1q9q2betyfU/1KGozZ860bLWRu3u4e3PWa/VDQ0OVnp6uYcOGWb75K3fCAQDAPJ+iHgBuP3PmzNHzzz8vwzAUGxurRo0ayTAMDRw4UO+8847b+1v12++NGzcqMTFRCxYs0K5du1SmTBlJUnZ2tl544QVLeqSkpCgmJkbR0dFq3769fv31V8drvXr1sqTH9YKDg50ChNTUVJdrdu7cWWvWrNHo0aO1Y8cOxcfHS5LCwsI0Z84cl+t7yo2+burWrasxY8ZYEiBIBYOK5ORkvffee8rNzdXTTz9tSY9r+2wcP37csc+GJMv22bh06ZL279+v7777TleuXFFQUJCkq2GVj48133I80aOoLV68uMT0mDZtmkJDQyVd3Zy1fv36+uc//6m4uDiNGDHC0vr/93//Z3l96eoKnZycHMcKnfT0dA0aNEilS5e29Ja9M2fO1OLFi2UYhmbNmqWnn35ar732mrKzsy3pYRiGli9frhUrVsgwDG3evFnjx49XYmKi7Ha7JT3czVNz2LBhgz766CMdPnzY6bgVmwnfTPfu3S2v6e55rF69WqdPn5Ykpaena+jQoWrdurUGDBig48ePu1x/woQJ2r59u8t1AHgWKxHgcR999JGSk5Pl7+/vdKlB//79FRcXp8GDB1vSx92rHapVq+Z4fP087rjjDrVv396SHpMmTdLixYsVFRWlTZs2qWfPnpo9e7ZiYmI8cilAr169LPnmnpOTo7vuukshISHavXu39u7dq1q1aunee++1YJQFXbhwQYcOHdKdd96p8uXLW1Lziy++sKTO77lZUHEtrLCCu/fZqFixoiZMmCDp6t+FU6dOKSIiQllZWfLzs+Zbjid6eEJUVJTjsc1mc3z+DcPQpUuXSkwPd2/Oen39n3/+WbNmzbK0vuSZO+F44pa9Y8eOVUZGhi5fvqyNGzfq7NmzeuSRR7R+/XodOnRIr776qgUzcS9PzGHKlCnavn276tevrzlz5qhfv37q16+fJGn+/Pnq2bOnyz2uBebS1RWF+fn5OnTokOO4Fbea9sQ8rr+D07UQz8o7OHniMj78MRs2bNCJEyfUqVMnlS1b1nH8448/tuRryjAMrVixQuXLl1eLFi309ddfa+3atYqMjNTTTz/tNb8M8HYl539b8Bo+Pj46deqU0w/hkpSWlmbZPxxz5szRsmXL1LlzZ8XGxurcuXM6ceKEBg4cqM6dO1sWVNzMU089pYULF7pcJy8vz/FDQLdu3RQREaFnn31WY8eOtWzzuPHjx0uSSpcu7fRDhWEYOnv2rMv133nnHS1cuFABAQF64YUX9NZbbyk2NlbTp09Xz5499dxzz7ncwxN3yri20uRGUlNTVatWLZd7eCKocPc+G0uXLnU8vj5cCwkJ0eeff+5y/d/rUb58ect6eEJwcLBWrVqlChUqFNi7pUmTJiWmh7s3Z71Wf/DgwWrRooXl9SXP3AnHE0HFtm3btG7dOl2+fFnR0dHauXOnAgIC1LVrV7Vp08aSHu7miTkkJSVpzZo18vPz0+TJk9WrVy8dOXJEEydOtCygr1atmsqVK6dhw4apcuXKysrKUteuXZWYmGhJfckz83B3iOeJy/hgnieCqWtBoc1m09KlS5Wbm1viwk4QIqAITJw4UT169FDNmjVVs2ZN5ebm6tixYzp8+LDjenlXeWK1w759+xyPQ0JCHMtRDcPQgQMHXK4vXV2enZaWpooVK0q6+lvFJUuWqE+fPjpy5IglPZYsWaJXXnlFYWFhOnfunNNry5Ytc7n+Z599puTkZF28eFHNmjXTN998o7CwMOXk5Khjx46WhAieuFPG77FqxUaZMmWcluvm5eXpP//5j6pVq+ZYzm0Fd+6zkZeXJ39/f0fIlZKSon379umee+5Rq1atXBv4Tbhj5YknPPbYYzp69KgqVKhQ4LWuXbuWmB7u3pz1Wv0WLVrI19dXs2fPtrS+5Jk74XgiqPD19ZV09XtHTEyMAgICJF39e15SfrvniTnk5+c7Vi2FhIQoMTFRL730kvr372/Z5yIxMVGrV6/WSy+9pBdffFHNmjWTv79/gV+guMIT83B3iOeJ/YZgnieCqWtBYfny5VWtWjXt3r27xIWdIERAEYiLi9PmzZu1e/du5eTk6MyZM6pcubIaN27s+M+Dqzyx2qF9+/Z64IEHZBiGAgIClJeX53jNit/gS1d/w56enu4IESSpSpUqWrp0qWW/zWjUqJGioqLUrl27AnexmDFjhsv1fX19FRgYqICAAAUGBjp+GL52HbvV3HWnjGsrNiTnVRtWrdiQpC+//FKjRo2Sj4+P5syZo8mTJyswMFCHDh3SlClTlJCQ4HKPlJQUDR06VHl5ebrvvvs0adIkVa9eXZLzXQ9uVfv27bV06VKFhIRo+vTp+uyzz9SqVSvNnTtXW7du1dixY12egydWnkhye6AzatSom742bty4EtPD3ZuzXqs/cuRI+fv769SpU5bWlzxzJ5wbBRXS1ZVMVgUVFStW1IULF1SmTBmtXLnS8W96Wlqa44fx4s4Tc6hRo4a2bNmiZs2aSbr6fWr69OmaNm2aVq1aZUkPSWrXrp1atmypt99+W++//77T/xOs4Il5uDvE88RlfDDPE8HU9UFh48aNS2TYCUIEFBEfHx/FxMS47faLnljtULt2bU2bNk13332325YJt2jR4obHy5cvr2HDhlnSY+7cuSpduvQNX9u6davL9evXr69BgwYpJydHcXFxGjZsmOLi4pSSkqLatWu7XF/yzJ0yrq3YCAgIULly5ZxWbVixYkO6Gtp89dVXunTpkhISErRy5UrVqlVLR48e1Z///GdLQgR377Nht9sdO9kvXbpUn3zyiQIDA5Wfn6+2bdtaEiJ4YuWJJwKdgwcPevwOJe5atWG322W32xUcHKy8vDzt27fP0sDlWqBTvnx5XbhwwfL6v/1cXL9CxyqeCCpudhld2bJltWDBAkt6uJsn5vD+++/f8PioUaMs28T2mqCgIL355pvauHGjdu7caWltT8zD3SGeJy7jg3meCKauBYXh4eFatGiR43hJCjtBiAAv5YnVDiNGjLjpTtFWBRXJycmKi4uTJJ05c0YjR47U3r17FRUVpQkTJtxwifIfdaP/hGdkZCg8PNzl2pL01ltvaeXKlbLZbOrTp4+++uorLVu2TJGRkerbt68lPTZu3Oh4HBoa6vgmZOWdMq6t2IiNjS0QGlmxYuOaa6tOqlev7thnoVq1apbtSu7ufTbKli2rH374Qffee6/CwsKUm5vrCBHcsTu8u1aeeCLQadOmje6880517txZzzzzjCV/n3/LE6s2rgUu15a/vv3225YGLp4IdDzxufjtpT7ffvut41Ifq0KEmwkKCtLx48ct+3fd3Y4dO6ayZcsqPDxcv/76q+Wb8QYGBrq9x2+DqXr16qlevXqW1L4mMDDQ6fn1IWHlypUt7SVdDfEuX76sxMREy75/e+oyPpjjiWDKG8JOECLAi7l7tUPHjh1v+ppV96qfOnWqI0QYNWqUKlasqMTERK1atUqjRo3S3//+d5d7XLt1k4+Pj06fPi3DMNShQwetWbNGhmG4/E3cz89PXbp0cTyOjY1VbGysy+O+nifulOHuFRvX2O12+fj4aO7cuY5jV65csWwZobv32Zg6daoGDx6sunXrqmLFio7Lfr7//nsNGTLE5fqSZ1aeSO4PdOrUqaPZs2dr2bJl6tat2/9j77zDorrWt/3MUAVEEJEiIiWKXWzYsECMNbEk9hxjrLGmGaMeC4qiqMGoQY0tevREYzTGFmNsYEmsiQ01UaQoIlKGDjLAzPcH396/GUCT47xryYzrvq5cR0bP+8zaM3uz17PfAisrK7mZmFRiYig8sjYkw6VatWpo06YNfvrpJ1LDhYehw+OzeF6pz8WLF5mnblP1bmENj2a8PDRMxSTU3VTa2toiLy8PkZGRKCoqAgCDjxUPk/B5UDVGftkaVEjG1PXr15GXl4f8/Hz4+PjgtddeIzOmnpWFZ2Njw6TUlVffJEpzzRgQJoJAYACxsbF4/Phxhc7auhkEVPz+++/4+eefAZQ1iNyzZw9J3GbNmsHDwwNKpVLeHKWkpKBXr15QKBQ4f/68QfFzc3MRGRmJ5ORk9O/fX2/sle4NUFXX4PFEZPny5VCr1bC2tkbbtm1lMyQ5ORlTpkwh0WDdZ6Nx48b45ZdfcPr0aTx58gQ+Pj5wc3NDSEgI2S/v8pknFhYWAGgzTwD2ho5CoUDDhg0xa9YsfPHFFzh+/DgOHDiAgQMHwt3dHQcPHiTRkWCVtQGUGS61atVCnTp1mBgurA0dHp/F35X6UJgIPHq3sIZHM14eGqZiEkZERCA4OBh+fn7QarUoKCiARqNBfn6+wbEBPibh8+BhrhmLgQcA58+fR2hoKOzt7RETE4PWrVsjOzsbFhYWWL16NerUqWOwhmSwDR8+HD169ECDBg0I3vn/YQrmmjEgTASB4AXZsmULtm3bhvr162PmzJkICQmRu8rqZhAYQnp6OjZs2CDfAGq1WjkVlqpL7ty5c3H27FlERETA1dUVANC+fXuyp+uffvopvL290bdvX3z//ffYvXs3IiMjYWVlpXcDVNU1ngfVSE9/f/9KX69bty7ZTSePPhtmZmYIDg5mlgXEI/OEh6FT/hxu2bIlWrZsiZCQELLzj1fWhrSZj4iIkF+jNFxYGzo8PgsepT48erewhkczXh4apmISRkVFYcGCBSgoKMCcOXNQUFCAPXv24NNPPyWJD7A3CXmYa6Zg4AFASEgIdu3aJU/sV0OAngAAIABJREFU+vjjj3HgwAGcOXMGn332mTya1hAkg+3YsWMYPXo0qlWrRmqwmYK5ZgwIE0EgeEF27tyJo0ePyg7k4MGD8fDhQ4wbN45sg//uu+/KF6SRI0dCpVLByckJqampZA3ZJk6ciH79+iEsLAxOTk7yaDMqEhISsGnTJgBlm+158+ZhyJAh2Lp1q1Fp8BjpmZOTg8jISBw9ehSZmZnQaDSoVasWevbsiSlTppA8yWetkZ+fj3Xr1uHIkSNISUmBubk56tWrh5EjR2Lo0KEGv39eGjwMnUmTJlX6ukKhkJtaGQqPfiGS4QKUbb4lqAwXHoYOj8+CR6kPr94tLOHRjJeHhqmYhHXq1MGmTZvwyy+/oE+fPkzStVmbhDzMNVMw8ICyz8LJyQkA4OnpiaSkJABlDyBCQkJINCSDLTAwEB9++CGuXr3KzGAzZnOtqiNMBIHgBSktLYWtrS0AwMvLC3v37sWECROQlJREZiLoXox0bwhr165NNh8dKEtn37VrF3bu3Ilhw4ahsLCQLLZarZZvEADgo48+gpubG9555x0yx5aHBo+RnhMnTkSnTp2wd+9eNG7cGOnp6UhNTcWePXvwwQcf4LvvvqvyGlOnTkWvXr3w7bff4tSpU0hLS0P//v2xevVqxMXFkaRs89DgYVQMHDiQJM7z4JG1wdpw4WHoVPZZqFQq0i70PEp9ePVuYQmPZrw8NF6GSciqtAsoS0EfOHAgZs+eTdq0kYdJyMNcMwUDDwCaN2+O6dOnIzAwENHR0fJ3tbCwkGwDztpgMxVzraqj0FLtdl6A5OTklyX9wpSO70caz2xTRbeNtQaPNfxTWKU789AYPHgwQkJC0LRpU1mjpKQEn376KX788Uc8fPiQVK/8Onbv3k22mdGNX1hYiMTERLIO1YsXL0aXLl3QpUsXvTVERUVh7ty5+PXXX41CIzg4GJs3b37mSM8rV64YrNG5c2ecPXsWQMXPW/fvqrJG9+7dceLECTl+QEAAjhw5Ao1Gg27duuHMmTMGxeelMXr0aPTq1QudO3euYFS4urqSGBU9e/ZE7969MWDAALRp04bJdYqnxujRo5k0reKxhiVLlmDixImoWbMmEhMTMWzYMCiVShQXF2P16tVkmz4J6dyjNip4auheP3hosEJoVB0NVvEzMzNhbW2NatWqGbWGLqw0iouL8e233+LevXsICAjAm2++CTMzMxQWFiIjI0PPmH5RfvzxRwwcOJDZGqTsCeD/MvAsLCygUqlw4cIFMgNdwsbGBrNnz8bVq1exb98+0thVAXd390pfV3J+H4JXhNLx/f7Rf08GdvxH/64qsnr1ar3mdEDZ9IE1a9ZwuYh88cUXTOJWq1ZNNhB2795tcLy5c+dWWocvpY5SwEODx0hPDw8PrFu3DmlpafJraWlpWLt27TMv4lVNw8bGBpcuXQIAHD58WG4kp1QqyTJ0eGg8fPgQQ4cOhbu7Oz7++GMcP34cPj4++PLLL+UGp4aSnZ2NnJwcDB48GJ06dcLGjRuRkpJCEvtlaPTs2RN9+/Yl1+CxhpMnT8qb4NmzZ2P9+vX49ddfsWvXLoSGhpJoLFmyBCqVCkBZo9wOHTrgrbfeQkBAgMENbF+WxvXr18k1oqKi5D9nZWVh+vTp6N69O6ZMmaJ33TImDWlEs7GvQ8pwoNTgsQZHR8cK4zCp4aHBAwsLC7z//vsICwvD2LFjYWZmhvT0dFSrVo3EQAAqz/yiNBM8PDzk/+rVqwcLCwukp6eTZuDpYmNjg3nz5umV47wKCBNBIHhB3N3dUbt2baSlpeHq1auIiYmRf+FRjTDs3r27/F/r1q3lP7/++uvMXW6AnVEhQWFS8NJ48803nzmiiWqk5/r165GZmYl33nkHrq6uaNKkCQYNGoSsrCx8/fXXRqERHh6OhQsXomHDhoiIiJA3YBkZGWTpfjw0eBgVNWrUwPz583H58mUsW7YM8fHx6NWrFwYNGkTSqJO3RmxsLObPn0+uwWMNxcXFcpprYWGhXELh6+urV7pkCDyMCt4aixYtItcIDw+X/zxr1ix5tHGLFi0wc+ZMg+O/DA3dEc3GvI5Zs2bB2dmZVIPHGngbOsaskZmZKf+nUqmgUqnQt29fZGVlySPBqTRUKpX8v5QaPNeguw5qjaqO6IkgMFr+aYbCk38Y738ty4iJicGsWbOQm5uLunXrori4GI8fP4a9vT3Cw8PRtGnT/yleZaSlpeHbb7+Fg4MDHB0d5QuTVqtF//79DY4PQB6HaGZmplfvptVqmRsVX3zxBWlJxsvSoCotcXBwwJw5czBnzhxmaX6sNRo3boyffvoJgH66pZOTE8aOHWs0GuHh4ZgxYwbu37+PZs2ayeOiKI0KXQIDA9GwYUMsXrwYZ86cwcGDB/Gvf/3L6DTatWuHdu3aMdNgtYb3338fI0eOxJQpU9CjRw/Mnz8fffr0wblz58ia2EpGhbm5OTOjgqcGADx9+pSJhgSr0cZCo2pqsIqvOzFL19A5cuQIZs6ciW+++UZo/H+ksd8A5NHflGO/dTVYjRbnuQaWGlUdYSIIBC/IJ598gmXLlqFVq1Z6G5nff/8dH3/8sVyzbQjdu3dHQUGB3HdBauQIgKxGVzIqvL299dxTKqNCMikAfaOC0qTgofE8jMmo4KERGxuLX375BVlZWSgqKoKLiwt69OhB1vWchwYPo8LHx6fCa2ZmZggKCiIZEWsqGjzWMGbMGDRs2BDbt2/HgwcPUFRUhLi4OPTs2ZNs9CkPo4Knxr///W8EBQWRa/AYbSw0qo4GjzXoYgpmC0sNaez33Llz0blzZ6Snp5OO/dbVYDVanOcaWGpUdYSJIBC8IAUFBWjVqlWF11u3bk023UB35np51q5dS6IhGRX16tXTMykAGqOCRzYFz4wNwHSNCgqNtWvXYv/+/ejfvz/atm2L3NxcPH78GJMnT0b//v0xdepUg98nDw2AvVGxfv16kjimrsFjDQDQsWNHdOzYkVkWEA+jgqfG5s2bcefOHZSWlpJq8BhtLDSqjgaPNZiC2cJLQxr7vWDBAhw8eBBTpkwhHfutq8FqtDjPNbDUqOqI6Qz/I2I6w9/HNyWN5zFv3jwkJCRg0KBBaNy4MbKyspCcnIy9e/fC09MTYWFhZO8tLS0NT58+RXZ2NlxcXODs7EwWW4LVjfP06dMxdOhQBAQEVNCYMmUKiRnCQ6NFixbPNSr++OMPgzWeZ1TExcUhPj6+ymsEBgYiKioKFhYWep+FWq0ma3TJQ0PXqGjQoIFsVBw4cMCojAqeGufOncP9+/ehUCjINUzlOEkYc/d2XvGFxqunwSq+7ohFGxsbDB48WDYqFi9eTDIy21Q0dLlw4QIWL16MpKQkXLt2jTQ2UPZ579y5E2vWrGGmwXoNvDReJs9qui1MhP8RU9gcm8IaeGn8HadOncIvv/wClUqFoqIiuLm5oUePHnj99ddJ3tPz+i4sXboUzZo1I9EBgNLSUty+fVveABiTUcFDw1SMCtYaXbp0wc6dO+Hh4aF3nJKSkjB8+HCSMZU8NEzFqOCpMWLECNjb2wMAqYapHCfAdMwQYRoJDWP7vCWM+T7kZWg8fPiQdOx3+fgsRouX12C5Bl4aL5NnmQiinEEgMIDg4GAEBwczu5g/r+/CJ598QtJ3QTIqCgoKZOOAhVGRlpaGhw8fMs2mYKnBs7SEZQ8M1hoLFy7E0KFD4e3tDW9vbxQVFeHRo0dISEggG4XJQ0OpVOLJkycVRlqlpqZCqaQZbLRr165KjYoJEyYgKCiIZOPKU8PNzU3vOkilYSrHyVRKfXRNI6mpIqWGqR0nofHy1wCYhtnyMjXMzMzINbZt26ZnRFJq8FoDa42qjMhE+B8xhSfsprAGXhr/lPImwn//+1+SjuGdOnWSn3iW19D9O0N44403sGzZMvTo0UMv/u+//46ZM2cabFTwyKbgmbHBo7QEMO4nGRqNBlevXkVBQQGys7Ph5uYGf39/mJmZGY1GVFQU5s6d+0yjgqKhn6lkbUga/v7+et8nKg1TOU6mUuojaZQ3jag0TO04CY2XvwZTyWYyNQ2RvWYciEwEgYAjVN5ccHAwRo4cWWnfhW7dupFosG4QySObgmfGBi+jwpizNpRKJTw9PfXMFkoDgYdGUFAQzp49y9SoMJWsDUmjQYMG8veIUsNUjhOP7BaeGm5ubkw0TO04CY2XGx8wnWwmU9MQ2WvGjTARBAIDeFZa2ciRI0niL1q0SO67cPjwYbnvwvvvv0/Wd0EyKsaMGQM7OzsAIDUqeEyx4KFhKkYFaw3d+J6enlCr1UzXwEoDMA2jgqdGfHw8/vzzT2i1WlINUzlOpmKGCNNIaBjb520KZospaggj0rgR5Qz/I6aQpm8Ka+Cl8Tx4pzKxTG8/deoUTp8+jYSEBHkDQNUgkscUCx4aPEtLKjMqKEpLeGiYwhoAfkYFwKdEhocG6+aspnCcTKHUR9JgZRpJ8U3lOAmNlx+fR3ma0PjfNZ5lRBqqYSrHqaogpjMQYQqbY1NYAy+N58Gjjk8XVn0XnhWfEtZTLHhomIpRwVrDFNYAmI5RwVOjoKAAtWvXhlarFdktz8EUzBBAmEZCw7g+b1MwW0xNQxiRxoHoiSAQEPOyU5l4+H9URgXrKRY8NHiWlrDsgcFawxTWAJhOiQxPjcqas1JomMpxMhUzRJhGQsPYPm/ANHr1mJqGl5cXCgoK9KYzUMY3heNUlREmgkDwgvCo4wPY9114HqyNCupsCtYapmBUsNYwhTUApmNUmIKGKawBMB0zRJhGQoNSQxh4r66GMCKNG2EiCAQvCI9GXDznJ5ef18vDqOCRTWFMGRuAaWRtmMIaTMWo4KnBqjmrqRwnUzFDhGkkNCg1hIH36moII9K4ESaCQGAASqUSrVu3ZraR4TFCRnder7+/PwCQGxU8silMOWMDML6sjZcRn1rDFIwKnhrR0dF6zVlFdos+pmKGCNNIaBjb520KZovQqDrxeWlUdURjxf8RU2gYaApr4KXxT2G1yejSpQt27twJDw8PPY2kpCQMHz4cZ8+eNVhDahBZfl4vVYNIHlMseE3KeJZRUb9+fZL4upT/Tu3YsYPcEGGtYQprqEyDhRnCMmuDl4YprIG1hik0mZU0WE30keKbynESGi8/vqlMiTI1jcqMSAoNUzlOVQUxnYEIU9gcm8IaeGn8U1jddPIYISMZFf7+/nproDIqeEyx4KFhSkYFaw1TWIMupmpU8JjwwlrDGI9TZRosEKaR0DA2DWHgvVoawog0DsR0BoHACOHRd0FqEPmseb2GwmOKBQ8NnqUlLHtgsNYwhTVImHqJjClomMIaAOMr9XlZGqawBqHxcuKbQq8eU9MYMmSI0a+BtUZVRpgIAkEVh3XfBcmoYDWvl8cUCx4apmJUsNYwhTUApmVU8NJg2ZzVVI7TszAVM0SYRkLDmOIDpmG2CI2qE5+XRlVAmAgCgQBKpRLt2rWDr68veWwe2RQ8MzaM3ahgrWEKawBMx6jgqcGqOaupHCfAdMwQYRoJDWP7vJ+FKZgtQqPqxOelURUQPRH+R0yh1t8U1sBL458iahFfDQ2NRsPUqODRA4O1himsAeDb1NTY+4Wwbs5qKsfJVJrM6ppG9vb2AECqYWrHSWi8/DUAptOrx5Q0zp07V8GIFL2Zqh6iJ4JAIBAYCK/SEmPO2jCFNQAi8+RFNNzc3JhomMpxMpVSH0mjvGkkSqKERlWMD5hONpOpaYjsNeNGmAgCwXP4p9kOT/5hPEOyHQSvBqyNCh4aprAGUzEqeGqwas5qKsfJVMwQYRoJDUoNYeC9uhrCiDRuhIkgEAgEAkElmIJRwVODVXNWUzlOpmKGCNNIaBjb520KZospaggj0rgRJoJAIBAIBC8JU8jakDRYNWeV4hv7cTIVM0SYRkLD2D5vUzBbTFFDGJHGjTARBAKBQCAQCDhgCmaIpCFMI6FhLPFNwWwxRQ1hRBo3wkQQCF4you+CQCAQCAQCATtMwWwxNQ1hRBo3wkQQCF4B/olRIUwKgUAgEAgEAoFA8HcIE0EgEBgMj2yKGWdmIC477m//vxYWFiguLv7bf+dTwwcruqz4h+9IIBAIBAKBQCAQAMJEEAgERkJcdhwupFxgqiFKSwQCgUAgEAgEgucjTASBQCDgCA+jglKDdXxT1xAIBAKBQCAwNV6NQZYCgUAgEAgEAoFAIBAIDEZkIggEAoFAwAhTyKgwhTWYkoZAIBAIBC+bf2QiXLt2DVu3boVGo8Hrr7+OAQMG6P29VqvF1q1bcfXqVVhZWWHy5Mnw8fFh8oYFAoFAIBAIXlV4NJllPdHHVAwdocE//rM0qsp5AVSdz+JZGgIBBX9rImg0GmzZsgVz586Fk5MTZs+ejTZt2sDDw0P+N1evXkVKSgrWrFmDe/fuYfPmzViyZAnTNy4QCAQCgUDwqsGjyaxAYGyI8+Kfw9pwMRWzRRg6z+dvTYTY2Fi4urrCxcUFANCxY0dcvnxZz0S4cuUKunTpAoVCgQYNGiA/Px+ZmZlwdHRk984FAoFAIBAIBAKBQPCPEYaLgIK/bayoUqng5OQk/+zk5ASVSlXh39SqVeu5/0YgEAgEAoFAIBAIBAKBcaPQarXa5/2D8+fP4/r165g4cSIA4MyZM4iNjcWYMWPkf7N06VIMHDgQDRs2BACEhobiX//6V4W+CCdOnMCJEycAAOHh4aQLEQgEAoFAIBAIBAKBQMCWv81EcHJyQkZGhvxzRkZGhTIFJycnpKenP/ffAED37t0RHh7+ShgIs2bNMnoNU1iD0Kg68YXGq6dhCmsQGlUnvtCoWhqmsAahUbU0TGENQqPqxDcljarI35oIvr6+ePz4MVJTU1FSUoLffvsNbdq00fs3bdq0wZkzZ6DVanH37l3Y2NiIfggCgUAgEAgEAoFAIBCYGH/bWNHMzAxjxoxBWFgYNBoNgoKCULduXRw7dgwA0KNHD7Rs2RJ//PEHPvzwQ1haWmLy5MnM37hAIBAIBAKBQCAQCAQCvpgtWLBgwd/9Izc3N/Tu3Rt9+vRBo0aNAJRlKPj6+gIAFAoFWrVqhT59+qBnz56oWbMm0zdtLJTvCWGMGqawBqFRdeILjVdPwxTWIDSqTnyhUbU0TGENQqNqaZjCGoRG1YlvShpVjb9trCgQCAQCgUAgEAgEAoFAAPyDnggCgUAgEAgEAoFAIBAIBIAwEQSCF0Kj0ej9XFRU9JLeiUAgEAgEAoHgWZS/ZxMIBIYjTASB4AVYu3at/EspNzcXK1eufMnv6NVk3bp1z/xPUJGsrCwAgFqtxvnz55Gdnc1EJyEhATExMfJ/gpdHeno6Nm7ciDVr1kCj0eDkyZPk8WNjY/VGQQsqJzc3l2l8jUaDX3/9FT///DM0Gg1iY2PJNbRaLXJyciAqYV8NtFotIiMjX/bbMJjFixdj48aNuH379st+KwbB2gxJSEjAihUrEBoaCo1Gg3379jHVM2aio6MBALGxsVi6dCkuXLjwct/QS+BvpzMI/hlqtRonTpzA3bt3UVBQgBo1aqBVq1bo0KGDwbH37NkDhUJR6d8NGjTI4PjliY+Ph7e3N3lcoOwX0o4dO/Dee+8xic+L4OBgREZGYsiQIdi0aRNGjx5NFnv+/Pny5617o6ZQKLBw4UIyHR6oVCr8+OOPSEpKglarhUKhgIeHBwYMGAAnJyeD4w8ePBgAcPjwYdSvXx8+Pj5ISEjAn3/+aXDs8vz222/o2LGj/PPBgwfRr18/ktgPHjzA3bt3kZeXBwcHBzRv3pxJg9o1a9Zg/vz52LVrF+zt7XH06FHy71R4eDhq1qwpj/lVKBRo2rQpWXwe15CoqChER0dDqfw/nz0kJISJlkajwfnz55GTk4OePXsiLi4Or732Gln89evXY8yYMdi8eTOUSiV+/fVXvP766wbHTUlJwfr16+Hi4gIHBweoVCpkZGRg4sSJcHFxIXjn/0dBQQGOHTuGvLw8jBgxAlevXkXr1q1JYvM694Cy82/OnDlMYkvxGzdujF9//RW9e/fGrl27MG/ePJLY2dnZ2LRpEwoLC2Fvb4+cnBzY2tpi3LhxsLe3J9HgTWJiIqKjo1FQUCD/rjW26WI5OTlMj79CoYCDgwPu3bsHHx8f+d5E99pIwbVr1+Dv74+UlBQcPnwYHTt2ROPGjcniz58/H8nJyfjtt9+wb98+eHp6IjAwkKQZHs9ryOLFi+Hq6orAwEDS4yOxdetWzJgxAxEREVAqlYiJicHbb79NFp/nsWLN2bNn0a1bNxw9ehQTJ07EsmXL0L59+5f9trgiTAQiIiMjERgYiNatWyMmJgYpKSkoKirC9u3bDb7ZbdKkCQDgzJkzcHFxgY+PD+Lj45GSkkLx1itw+vRp7NixA40aNUKnTp3g7u5OFluhUECtViMzM1PeZPDgm2++wZgxYwyO8+TJEwBArVq10LRpUyxatAiTJ0+GhYWFwbElQkNDyWI9iwcPHuC7775DYWGhvLm3sbHBkCFDUK9ePTKdyMhIjBgxQm9TFBsbi7Vr12L+/PkGx3d2dgZQ5p5LRo67uzt++eUXg2OX56+//gIAdOzYETt27ED16tVJ4n777bdQq9WoV68eHj58iJSUFNy9exd+fn7o2rUriYZEcXExgLInoqNGjcK1a9dI40saEyZMII8rweMacvz4cYSGhsLcnP2vSJYbP6DMpKhTp478M9UT5G+//Rbjx4+Hh4eH/FpSUhJ27NiBzz77jERDYs2aNejVqxd+/PFHKJVKHDlyhMRE4HnuAYCjoyP2798PHx8feRNGabDl5uaiR48eOH/+PFlMiR07dqBXr1567/f27dvYvn07pk6dSqbDc5Oxdu1ajB07ljw+zzVs2rQJpaWlCAgIQLt27VCtWjVyjfv37+P+/ft6r1GbqocOHYK/vz/27duH7t27Y+vWrVi6dCmphpmZGZRKJRQKBZRKJX777TccPHgQH3/88QvH5H0NYWmGSNjZ2cl/Li0tJYvL61jxOv+ePn2K27dvw9raGo6OjrCysiLXqOoIE4GIrKwsBAQEACjb2CxatAjvvvsuFi1aZHBsyW3ct28fJk6cCABo0aIFSezKeP/996HVanHnzh3s3LkTaWlpCAwMRFBQkN7F5UWJj49HSEgIatSoAYVCQfqEfffu3RVe02q1uHr1Kkn8H374Qe/nJk2a4PTp0wDon2BkZGTg+PHjyMzMJH9KsnnzZnzyySd6mzCVSoVVq1aRmhhqtRp169bVe83DwwNqtZpMAyg7R5YuXQpPT088ePBAHkVLyejRo7Ft2zacOnUKnTp1QlBQEEnc+/fvy4ZKcHAwFi1ahHnz5mHRokXkNyGNGjXCwoUL0a9fP6jValhaWpLGB4CgoCBs27YNnp6e8mvBwcGkGiyvIUDZxu7Ro0d6313qJ28SLDd+QNk1auPGjcjMzMTWrVvRrFkzkriFhYV6BgJQdm4/ffqUJL4uxcXF8Pf3x4EDBwDQGSE8zz2g7N6guLhYNiSps3Ts7e3x22+/Qa1W49KlS3BwcCCLnZWVVeG9Nm7cmDTdmfeGzNPTE76+vqRmIe81TJ8+HYWFhbh06RJWr14Nc3NzBAYGIiAggOyaFRISgpKSEuTk5DAzQwoLC5Geng6lUokGDRqQb8gWL14MZ2dndO7cGQMHDpQzKr7//nuD4vK+hgBszBCJrl27Yvny5UhNTcXKlSvJ7nMAPseK5/k3atQoXL16FYMGDYJarUa7du1I4xsDwkQgolWrVggPD4eHhwfu37+Pzp07AwBsbW3JNJycnLBx40Z4eXkhISGBJB28MgoKCnDhwgVcuXIFNjY2GDp0KBQKBVasWEFyox4WFkbwLivn1KlT+Oijjyq8TvXEVdrEFxQUwMbGRn69oKCAJL4ua9aswaBBg7B7924MGTKEzAiR4FHTOmzYMISHh8PKygrW1tYoLCxEcXExhg4dSqozdOhQZGVlIS0tDX379iW9eS5fXvL48WOo1WpER0eTnA/29vbYv38/6tWrh1u3bskbMxa1jyNGjND7mUVq9bFjx9CyZUtmm26A7TUEAO7du4d79+7pvcaqnIHlxg8oK3l78OABmjVrhjp16uiZO4ZQWFgoZ2ZJaLVaFBYWksTXxc3NDfv370deXh4OHz5cwbx4UXiee0BZ+dXdu3eRlZWFNm3akPeRmDRpEk6dOgVvb29kZGTggw8+IItdWloKrVZb4fcG5ZNK3huyxMRETJ48Ga6urgBoSgZ5r0Gr1eL+/ftyCV+zZs2Qk5ODpUuXkl3fo6OjcfbsWeTk5GDZsmVYtWoVPv30U5LYEgMGDMB3332Ht99+G2q1GvXr1yeN37lzZ73jf+nSJQQEBGDIkCEGxeV9DWFlhkgEBwejbdu2ePLkCWrXrk1aKsPjWPE8/+rVq4ecnBzcuHEDWq2WdL9nLAgTgYgBAwYgKysL6enp6Nevn3ziUV5oJ02ahNjYWKSmpiI4OJi0blaXyMhIdOzYER9++CGsra3l16meHl+/fh0HDx5EVlYWVqxYgW3btpGUGgBAz5494eHhUeHC16NHD5L4EhEREXopx+V/pkCpVKJZs2bYu3cvmjdvLj+Fo2D8+PH45ptvkJ+fD41GA4VCATs7O4wbN45MAyh7otu0aVOo1Wrk5+fD1taW9Ol3VFQUgoKC8N1331XoG0JlVLAuL/nwww9x6dIlPHz4EA0bNpTTtKdNm0aupXuDnJmZiRo1apD3RLCzs8PAgQNJY5aH5TUEAD755BNudd6sNn6nTp2q8FpsbCxiY2NJMkPq1KlTITNLep2acePG4ffff0fnzp3h6uqKN998kyQuz3MPALZs2QJra2vExMQgICAAGzZswNy5cw2Oq2vmtGzZEi1btgRQdo5T9qdgfS3kvSFbsWIFeUzeawgNDUWzZs0wcOBA1K6oUmblAAAgAElEQVRdm4nGqVOnEBoaioULF0KpVDJpENqmTRsUFxfj+vXr6NmzJ+lTXY1Gg+joaHTp0gVarVZuMCtlDxsCz2uIVqtFw4YNK+2FZqgZUtk9lATVvRSPY8Xz/AsLC0PLli25lmZXNYSJQIiDg4PeU6Tdu3eTPnFVqVS4d+8e8vPzkZycjGvXrjFprCg1bZH44Ycf8M4775Bd1Pfu3YuQkBCEhYVBqVTi4cOHJHEBPLMBDHU6dXlDhTo9HyhzOdVqNZo2bYqFCxfqGTqGUrduXfK65cp4+vQpjh8/Ln9vbWxsUL9+fbzxxhsktZtSA9DmzZsbHOvv2LVrF4YPHw6g7Jf57t27MWzYMIPjKpVKuLu7Izc3F0lJScjLy2NWw6f7ND0nJwd79+4l15CylurWrSvflFBnnrC8hgBl9d85OTmoUaMGGjRogIYNG5I9/ZZgvfFjmQkClNX3S+cDa7755hu0b98eb7311jNvdF8EnuceADx69Ajz58+XjTuqG9vKzBwJqhK4tm3bok+fPiSxngVvU2fJkiWoVasWGjRoAD8/P7i5uRkck/em0tvbu9L7HsoHJ2ZmZnKZklqtZnJtYdUbJjo6GqdPn0ZCQoJsgpmbm8vXW0PheQ1RKBRISUlhUorI4x6Kx7Hief7Z2toyf2BS1REmAhGTJk1CrVq1oFQq5XS/hw8f4vbt22RP+lasWIG+ffuSNr7TJTU1FU+ePNEby1ZaWoqbN2/inXfeIdNRKpXyLyHWKfV79uyRO/hT0qRJE6xZswZ+fn64e/eu3PySktdffx2WlpYYPHgw8vLymI3jA9gdp9WrV6Nz584ICgqCjY0NCgoKcPPmTaxevRqzZs0yOH5eXh63EYJ3796V/6xQKOS6ZkPhWcOnu2nRrc2mpG/fvuQxy8P6GjJlyhQ8ffoUly9fxqFDh/DDDz9g/fr1pBqsN37dunWT/1xcXEx+/dA9H1jTvXt3XLx4ET/88APc3NzQvn17kl4CvOvXra2t5TKZ+Ph4vZI4Q9D9vki16zVq1ICZmRlJfAC4fPkycxOBt6nz73//GyqVCjdu3MCXX34JlUqFzZs3GxST96aSR6Pqd999FxEREXj06BEiIiKYmIesesN069YN3bp1Q2xsLJPsXd7XkMTEREyaNAnu7u6k/YCk3mulpaW4ePEiUlJS4OrqSpKtIcHjWPE+/1g/MKnqCBOBiFGjRuHixYto0aIFOnfuDDMzMyxZsgT//ve/yTScnZ31MgSoSU9Px59//on8/HzcuXMHQJkDTf0Lo3///li0aBGSkpIQFhZGNiavMljNBB42bBgSEhKQnJyMt956C15eXuQa33zzjfzk2M7ODhs3biSvQ5RgdZzy8vLQvn17ecNnZ2eHdu3a4ciRIyTxpe9peagblgGApaUlbty4IRtHVM24eNbw6TZjtbW1JcmkKI+Xl5c8jm/48OFMJkCwvobMnDkTjo6OCAwMxIwZM+QpIJTwGiW3f/9+XL9+HcnJyXBxcYGFhQXJE76kpKQK8+OlSS+U3fqBsgZ4np6eyM3NxYEDB7Bs2TLs2LHD4Li869cnTpyI/fv3w9LSEmfPniWfYnLy5ElER0fD2dkZaWlp6Nq1K7p3704SOzU1tdLGxQDdjTPvDdn27duRl5cHKysr9OnTBw0bNjQ4Ju81sG4yC5Tde7IcTQqw6w2zdetWjB49Gtu3b9frbUR1nHhfQ1iU4OiycuVK+Pn5wdfXF4mJiVi5ciU+//xzktim1liRqqzOmBEmAhHt27dH+/bt8ccff+Crr75CgwYNSBsOAWUbspkzZ8pptdQ3a40aNULjxo0xcOBApqmwrVq1QsuWLZGbm8uk7njPnj1QKBTQarVIS0uTU7YpSz+0Wi1yc3NRUlKCxMREJCYmkl2goqKiEBUVhQcPHiAkJET+hUfdSJPHcerZsycWLlwIT09PVKtWDQUFBUhKSiJLtWSRPfEsJk2ahP379+Onn36Cu7s7pkyZQhKXZw1fSEgISktLkZ2dDQcHB2ZpqdI4PjMzM7JxfLroXkOqV69OmuIOlD15++uvv3Dt2jXExMTA09OT2VPYGzdu4NChQ8jIyJB7k1BuAq5cuYLFixdjwYIFWLBgAb788kuSuLVr1+b21OXYsWP4448/YG5uLvcSoIB3/bq9vT0GDRokN+KlLoOLiopCaGgolEolNBoN5s2bR2Yi2Nrakk32eBa8N2QWFhYoLS1FcXExnj59SjJZhPcaWDeZBcp6eajVavj7+6Ndu3ZMsh7K94aRJpEZijT6mVU/D97XkIyMDPz444948uQJXFxcMGDAANSqVYssvlqtlk35Fi1a4ObNm2SxTa2xYsOGDeWsDRcXF7Rv3540vjEgTARiWrVqhVatWiEmJkbvxvbx48cG19tNmjTJ0Lf3XLZv345Ro0ZV+kuJsjP59evXcfz4cRQUFMipyJTxddN4r127hm7dupGnPH/55Zdwd3fHxYsX0a5dO/mpDwVBQUEICgrCiRMnyG4AK4PHcQoMDESHDh3w+PFjeaKFm5sbaZotAJw5cwYnTpxAUlISbGxsYGdnh/DwcFINBwcHvP/++6QxAb41fOfOncPPP/+M2rVrIzU1FT179kSXLl1INViN49Nl/fr1mDRpEuzt7aHVavH111+T3XQCQM2aNeHo6IjU1FRkZGQwuyEEINf+Ll++HJ9//jl27dpFGl/KmLGyssLt27eRlJREEtfS0pJJhkZl2NjY4OOPPybtCwPwr8HfsGED0tPT9TZhFBkp0vfTz88PsbGx8PLyQmJiIvz8/AyOLWFrayunPLOC94Zs+PDhSE5Oxu3bt3Hs2DEcOnQIa9euNSgm7zU8ePAAO3fuRGFhIWxsbDBs2DDyktdPP/0UarUaV69exY4dO6BSqbBgwQJSjd27d2PkyJHyz1IvLkNh3TCQ9zVk3bp1GDx4MHx9fXH//n2sXbuW5P5ZasRrY2MjT4FLTEwk6V0lYWqNFSMjI+Hh4QEfHx/ExcVhzZo1JGM2jQlhIjBC6kovsXHjRoNP9Fq1aiEmJgaZmZnyjTmlszZq1CgA7EaZSezcuRMzZ85kVueoe2NraWlJ6tJK5OTk4NNPP8Wff/6JYcOGYdmyZeQavr6+WLFiBQoLCzF37lzs37//mY0jXwQex0mj0eDy5ct6jRUbNGiAtm3bkhoJP//8M8LCwhAaGop58+bh66+/JostUd6osLW1JfncedbwHT16FKGhoTAzM0NJSQlCQkLITQRW4/h0SU1Nlf+sUCgqjBo0lOjoaPj5+WHkyJHMpzRYW1vDxsYGCoUCFhYWuH//Pmn8999/H8XFxRg5ciSOHTumd6NuCOPHj9f7+fr169BqtWjRogVZZog0fSUpKQmPHj3S+zuKDQDvGvzU1FTyKT6AfplSXFwceXygYtZXcnIytFot6TQO3huypUuXws3NDQ0bNsSsWbNIPnfea9i8eTOmTZsGZ2dnpKam4quvvtL7PlARFxeHO3fuQKVSkfYWkHpx3bx5k0kvLtYNA3lfQ9RqtVx207BhQ5SUlJDElbISdRtOUo/ZNLXGihkZGfjwww8BAP7+/sz3TlURYSIYESyffusSFRWF6OhovVRnypPDz88PFhYWZPGeB6vu4WZmZiguLkbNmjWxd+9eqFQqco1t27ZhxowZiIiIgFKpRExMDKmJoAur47R27Vp4enqiU6dOsLGxQWFhIW7evIm1a9fKF18KrKysoFQqYWZmhszMTCQkJJDFlmBlVPCs4VMoFMjMzEStWrWQlZXFpJyB1Tg+XapXr46TJ0/K/Sns7OxI448YMQIXLlzAyZMn5TRFViVeXbt2hVqtRu/evRESEoJWrVqRxE1PTwdQ1ockOzsb1tbWpL0jNmzYgNmzZ8Pa2hqbNm0CUPYU6NdffyUr9WE9fYXXuSdtjpydnXHkyBG9RlwUvVt43LwePnwYXl5esLGxwd69exEbGws7Ozs4ODjgX//6F4kG7w3Z7NmzkZmZibS0NDLji/caSktL5VJHJycnJk9cFyxYAD8/P3Tp0oU8G0+3F9eff/4JrVZL2ourWrVq8Pb2ZtaAmXcPjDZt2mDJkiXw9PREYmIiWakg60a8gOk1VqxZsyb27dsHb29v3L9//5Uc9ShMBCOCx9NvoKz+dNGiRWSN4yTmz58PhUKB0tJSzJ07FzVq1AAAJo2AgLI0v9LSUrlpIGUq5r///W+YmZlhwoQJuH79OvkISQndzRF1jw0JlscpLS2tggPs7e0t16xRMWDAAKjVagwaNAhbtmxhUr/OyqjgWcM3btw4fPPNN8jPz4etrS3Gjh1LGh8AvvrqK0ybNk2+uZEaW1EyZcoUnDhxAkePHoW7uzt5Iz+eaYrSzVtAQABpJ+wpU6bA19e30kwQihR6hUIBa2trqNVq3LhxA1999RUAkF7LpYa1J0+e1LuObN26leQ6xevckxrAOjk5yZslgL4BbEJCAg4ePIisrCzyUkGpHK20tBSnT5/G6tWroVQqSQ0M3huygwcPIiYmBl5eXoiPj0eTJk0wYMAAg2LyXkOvXr0wd+5cuZlmr169yDWmTp2KrKws+Z6NksaNG8u9uOLi4pCVlYU2bdogIyODJH5iYiK8vb0rNGGmOvd498AIDg5G165dkZaWhjfffJP8gRyrRryA6TVWnDZtGi5duoTExER4eHi8kuMehYnACYqZrjyefgNAs2bN8OjRI9StW1d+jeIpnNTY5sGDB/D09JRfT05ONjh2ecLDw+W6ZqDsFwbl5vjp06c4fvy43NyGxYjHrl27Yvny5UhNTcXKlSsRFBRErsH6OLVp0wbh4eFo3LixPOLx9u3b5I32pIZYjRo1QqNGjUhjS7AyKnRr+G7fvs20hq+0tJSs03J54uPjER8fj7i4OLm+UqPRMMkKsbKyQrNmzeQn1ffv3yf93vJMU4yOjsbJkyf1rrEUG/GwsDBcvnwZDx8+hIeHB9q1awdfX1+D40potVqkpqbi+vXreg33KJsFJiQkIC4ujtl3ite5J5UCHDp0CG+99Zb8+okTJ0h1Nm3ahGnTpmHDhg344IMPEBUVRRa7pKQERUVFiImJQf369eXvq7E2RQPKmo7qNtybN2+ewSYC7zV4eXlh8eLFyMnJgb29PVJSUship6amYsOGDXB0dETNmjWhUqmQmZmJCRMmwMXFhUwHAP7zn//A2toaMTExcvPUuXPnGhxXMmkHDx4MlUqFtLQ0ODs7kz2Z5vn7GyibnhASEiJPr1i5ciXp1C5WjXgBPseKx/kXHx8Pb29v3L59G3Z2dnJ5z507d8inglV1hIlAyMOHD6FUKvVqBO/du4f69etj9uzZBsfn9fT73r178ixrCcqb6K1bt+rF++6778hHF2q1WvLxWbqsXLkS3bt3R0BAABITExEREUG+0QgODkbbtm3x5MkT1K5dm0ltNuvj1K9fP3lGc0FBAWrWrIlu3bqRryUzMxMRERGoVq0aAgIC0KpVK/ImbP7+/gCA1157DdOmTSOb8S7V8D148AANGjRg3lhxx44d8PHxQfv27UlrHhUKBZRKJezs7OQNhqWlJZNRhqzNL55pir/88guTzK/XXntNvrm5d+8eNmzYAB8fH9Ku5//9739hZWUlp7OnpKTI5wkVLL9TvM69/Px85Obm4sKFC3K2iUajwfnz50mb51pYWMDV1RUajQaurq5yxgMFQ4cOxdKlS6FQKPDBBx8AKPu8KXue8N6QmZub46+//pLPc4pzkPcapFHQ0qaS8n5qy5YtGD9+PFxdXeXXHj9+jC1btpCOLweAR48eYf78+bKBSn28vv/+e8THx6Nu3bp4+PAhvL29MWTIEIPj8rqGXLp0CRcvXtQbrVtaWorCwkJSHVaNeAE+x4rH+cc6u8WYECYCEdu3b0dWVhbMzMyQm5uLyZMnw97eHjt37iTZXCYkJKBatWpwcXHB5cuXkZeXR5LdUB6pMZahbnxl8BpdCJR1mN22bZtexgOl6WJubo4OHToAANzd3Umf+LDuJqwL6+MElF3Uy9d5x8bGkjZn6t27N3r37o2cnBwcPHgQGzZswH/+8x+S2BcvXsSRI0dgZWWFjh074uzZs1AoFGjUqBFJ4yelUlnpaCAWNXzvvfcegLImWdHR0Vi1apXB3cglvLy84OXlhXr16sHb2xvZ2dk4ffo0WeMnXVibX+XTFFn1IgHKmlexyPzKycnBlStXcPPmTVhYWODtt9/Wa5plKF5eXhU2K66urqRjV1l/p3ide7dv38bly5eRlpaGH374AUBZZuEbb7xBquPv7w+1Wo3AwEDMmDGD9BrbrFmzCiMeXV1dZUOBAp6GKlBW1nPgwAH88MMPcHFxIenlwWsNPO6n1Gq1noEAlDXOpR5NCpQ1mJUeXsXHx5OZ9BK3bt3Sy/CieujD6xrStGlTeHt7o1atWrLxaGZmJptHVLBqxAvwOVY8zj8pu8Xa2pppZpkxIEwEIu7fvy9foBITE7Fy5UqyZkORkZGwtLREYWEhcnJy0LZtW9ja2mL16tWYM2cOiYaEQqFAUlIS1Go1uUnBa3QhUDbflgXSBl+pVCIsLEweg0M5aYB1N2FdWB0niWe5v9JYOyqSk5Nx/vx53Lp1C87OzqT16wcPHsSCBQtQVFSEGTNmYO3atVAqlZg/fz6JiVAZ2dnZ+O677xAYGEhaKlNYWIgrV67g0qVLUKvV6N+/P1lsiR07dmD+/PnYvXs3GjdujHXr1pHPMmdtfhUXF8Pc3ByOjo4oKirC2bNnmaQiA2WjQ7/88ku9bAeKG9wPPvgAXl5eaNKkCczNzZGYmIjExEQA9GYka3h8pyRYnHtt27ZF27ZtkZ6eDktLS1SvXp2skZ8u/fv3h0KhwBtvvIE33ngDxcXF5BosqWyTkZ2djb1796JTp07kZYO1atXC8OHDUVBQQBaT1xp43E/Z2dnh7t27aNCggfzaX3/9Rd7IFgAmTpyI/fv3w9LSEmfPniUziaXJPXXr1sW5c+fg7e2NxMREPdOWGhbXEBsbG9jY2GDo0KG4desWsyltGo0GFhYW8PDwwJgxY3Dz5k2y2JVBfax4nH+6mWXt2rWDVqtlkllmDAgTgQiNRoOSkhKYm5ujXr16+Oyzz/DVV1+RpAJlZGTIN5UzZ86UG+dQPv3WJTExEZMmTYK7uzsUCgVp48Pr16/j3r17uHz5MhwdHdG+fXvy9Feg7Cmcn58ffH19SdOEpQ2+7hMZyqd7wP81NiwoKMCxY8eQl5eH4cOH49q1a6Q6ALvjJPHee+9VmjIvbWaoOHToEDp06ICBAweSd9G3tLSEhYWFnCosxac0jspTvXp19O/fn/w4rVmzBq1bt8b48eOZjS4sKipCcXExSkpKEBgYiOPHj5NrsDa/wsLC0LJlSy7dlq9du4ZVq1aRx2UxSvBlweM7JcHi3MvNzcXWrVuRnZ2NGjVqIDs7G/b29hg9ejTpeRgWFoaRI0fKqbx79+4lb2LLm+rVq6Nfv37k10KgbMJIenq63nnOovyK5RoCAwNx6tQpvR5ZgwYNIok9fvx4bN26Fbt27YK9vT2ys7Ph4ODAJAvM3t4eb775JlJTU0nLN6XMHwC4ceMGbty4QRL3ebD6/Q0Aq1atYjql7aeffkLNmjXRp08f7N69G9bW1hUykChheax0NSjPP93Msr179wJgk1lmDAgTgYhRo0YhPz9f7l5rZ2eHmTNn4vz58wbHVqvVSE1NhUajgbm5ufxnFillALBixQomcQ8cOIDU1FS8/fbbcHJyQkZGBg4cOICEhATy8gkfHx/cunULhw4dQklJCerUqUOSlkVZd/13rFmzBr169cKPP/4IMzMzHDlyhLwhIavjJFGnTh3MmDGjQmoi9RzrUaNGISYmBufOnSN35+Pj4+V0Ud3U0YcPH5LErwylUom8vDy0a9eOLKZWq4WNjQ1zp7xLly5Yvnw5hgwZArVajdq1a5NrdOvWjdkYKgCwtbXl1mnZx8cHly9f1hv7R9G0jPW1as+ePc98mk61iZHo3Lkzs+/UgwcPcPfuXeTl5cHBwQHNmzeHq6trhTRuQ9i0aRP69u0LPz8/+bU///wTGzduxGeffUamM3XqVGzevBlqtRrOzs6YMWMGWWwJlUqFixcvIj8/X36N+vPW5dixY+jVqxfp5yGRmprKxWxjuYaIiAh06NABly9fRo8ePXD//n2y2Pb29vjoo4+g0Wjkxo2sRt2W71ng5eVFkjHFwhQqT1xcHJycnFC9enX8/vvvsLS0RIsWLZh83qyntE2bNg0bN27E5MmTMXr0aPTo0YM0fvnecUqlErm5uaT3Oqyv6bqZZbVq1SKJaawIE4GIymoPlUql3k3Di1KnTh3Z7Sr/ZxYkJCRgz549KCwsxNy5c7F//36SmuCrV69iwYIF8s8uLi6YMGECQkJCyE2E6tWrw9bWFtWrV0dJSQn5GBweFBcXw9/fHwcOHAAAeXNMCevjNGvWrErLYqibMi1ZsoTZk+Nt27Y99+/v3Llj0EQIXiUfCoUCNWvWRGxsLLy9veUNIPVNoaOjI2bOnClntlDUGZeH5RgqoOxYrVixQm9jz6oEoKCgAJcvX8bly5fl13jc+BqKlBZ65swZuLi4wMfHB/Hx8aTd4SV69eoFHx8fZGZmwtfXl6QhGsBvHFhubm6Fe4GGDRti9+7dZBpA2WYmPz8fzZo1w507d5CRkUHa+BAoe8jQt29f1KtXjzQu8H9joIH/+3338OFDnD9/nskYaGdnZxw5ckTvPDe0MRrvNZSUlCA4OBinT5/G66+/jkuXLpFrKJVK8tr78rDqWSBx48YNHDp0CBkZGVAoFLC1tdWbzPGirF+/HlqtFhYWFsjJyYGjoyNsbGxw4cIF0n4hEqyntK1Zswa1a9dGZGQkvv32W6hUKgwbNowkNuvecQCfa7o0tnrNmjV65zqrcfVVGWEiMObrr782eEyN7g2lSqWCSqWCr68vs0yErVu3YsaMGYiIiIBSqURMTAyJiaDVauUTTUKj0TDZHC9duhSenp7o1asXGjVqhGrVqpFrsMbNzQ379+9HXl4eDh8+TH4zCLA/Ts/a1FOXAvB8clye77//3qBfgLxKPoCyhpaxsbF6r1HfrKWnp2PZsmVwdHREp06d0KxZM3KjguUYKgB48803SeM9j8mTJ+td14uKirhpG4KU6bBv3z554kOLFi3Is4yAsi7xLMa/8RrHZ2lpKY8DlkhJSSEvIbt16xZmz54NS0tLZGZmYtu2bfjkk09INZydnREYGEgaU0KadtStWzfZpFqyZAm56Szh5OSE/Px8eYoFRXd13muwt7eHWq2Gp6cn1q1bR96tnzW8ehZIpvzy5cvx+eefY9euXSRxU1JS5I3j9OnTMX36dAA0Y3org/WUtn79+sHLywtAWZNCXXPbUFj2jtPVYH1NHz16NACQmFDGjjARiJg6dWqFrrjUKc/79u3Dw4cPkZycjPDwcERERDD7xaTbOKe0tJQkZnBwMFasWIHevXvD0dERKpUKR44cYTKq8uuvv0ZaWhpiYmKwcOFCaDQaLF++nCx+VFQUoqOj9TZH1JuxcePG4ffff0fnzp3h6uqKPn36kMYH2B8nXvB8ckwNr5IPoOw7WlJSgpycHCbTHwCgT58+6NOnD9LS0rBjxw5ERkaiQ4cO6NOnD1k6IcsxVACQlZWFjh07yj8fPHiQWXlA+ev6ypUrSa/rUVFRCAoKeubPhuLk5ISNGzfCy8sLCQkJTKbtsBr/xmsc39ixY7FhwwY4ODjIv/uysrLIn1TqlqI5OjqSGghfffUVFAoF8vLyMHPmTHh4eMjX2qlTp5JovPnmmygpKcHJkydx/PhxZmaFRMeOHfUyOsuPbHsReK5Bq9Xi7bffhqWlJUaPHo2EhAS4u7uTxU9PT3/m31GlcfPqWWBtbQ0bGxsoFApYWFiQlX3oXiuGDx8u/5n64dipU6cqff3atWuk99D16tXDzZs39Ro3UsGyd5wEj2v6unXrnvl3xpBFSIkwEYiwtrbGnDlzKqRuU24Ebt68iZCQECxcuBAKhYLJ6DSgrJZ8+fLlSE1NxcqVK8luOLt27QpfX19cuHABmZmZcHR0xIgRI/Q6rFOxY8cOJCUlwcLCAgEBASRlJbocP34coaGhTJoRSmzbtg3vv/8+WrduDbVajdWrV5M/VWJ9nHjB88kxNbxKPgAgOjoaZ8+eRU5ODpYtW4ZVq1aRzRSXSEpKwrlz53Dv3j289tprcur52rVrya6HLMdQAWUdyIGyTcaOHTtQvXp10vi6sL6ue3t7P/dnQ5k0aRJiY2ORmpqK4OBg0rGCEqzGv/Eax1e7dm3MmzcP6enp8u8+3U2YoSVREmfOnMGJEyeQlJSEatWqoXr16ggPDzc4LgCylOa/w9zcHD179kT37t1x5swZJmUTEt9++y3effdd1KlTB5cvX8bZs2dJPgdea1AoFNi1axdmzZoFpVIJHx8f0vjff/89gLIM2MzMTHh6euLBgwdwcHAgKx/jtenq2rUr1Go1evfujZCQkApjp1+UCRMmQKPRQKlUok2bNgDKSkyo70lY9aIoz5dffsmscSPL3nESPK7p0hjjw4cPo379+nIpn3Tf8CohTAQipk6dWqnTRXmBNDc3l+ufsrKymNX5S01Dnjx5gtq1a5Omnnt4eGDQoEFITExEQUEB8vLycPv2bfKnfEFBQUzS/yWaNm3KZLa7Lq+99hq2bduGIUOGYM2aNejXrx9pfID9cQLKOqsfP35c/qUaFxeH7OxskqkWMTExBscwFGdnZ4P+/7xKPoCypxmhoaFYuHCh3NCImqNHjyIwMLDCpuPdd981OHZcXByUSqWcbunh4YHg4GDyJ8ejR4/Gtm3bcOrUKXTq1In0yX15WF/XpWMlQX2+p6en49SpU3j69CmmTp2KkydP4m1j+5MAACAASURBVPXXXyfVYDX+jdeMd4latWpV+gTX0JIoiZ9//hlhYWEIDQ3FvHnz8PXXXxscU0K6zknpwRKrVq0iHakrYWZmJp93sbGxTMypadOmITIyEl5eXsjIyCBfh+4aWGFtbY1NmzbBx8dHzgyhejIt3b9+8cUXWLFiBZRKJTQaDb744guS+Lqw6lkg0b59e1haWiIgIAABAQHIyckhiVtZ2YW5ublsKFDRrVs3AM/PDqGAZePGZ/WO69SpE5nGs67pKpWK7LouXQsTEhLk0gZ3d3ccO3aMJL4xIUwEIsrfqElIqZ1Lly7F7NmzDdIYO3Ystm/fjvz8fGzbtk3+8lKzcuVKhISEyE/fVq5cSfq0Mjw8HDVr1pQ3TwqFgsxE0Gq1OHHiBC5duoTS0lKYmZmhbdu26N69O+km/969e/KTMQmqcgZpQ9SxY0dotVrMmzcPU6ZMeeZ37EXgdZyAspTzW7duoVu3brCzs8ORI0dkJ9dQ7ty5A4VCUSHtjqK2tTyhoaFo3rw5OnbsqNcd3pjS18zMzPD06VMAZVNfKD9rlUoFe3t7jBs3Dmq1GidOnIBGo0GXLl1gbW2Nhg0bGqzx3//+F59//rnea66urli2bBnJ+Ve+Kdrjx4+hVqsRHR3NrMaV13VdIjw8nKSfgMT69esxZswYbN68GUqlEr/++iu5iWBvb4/33nuPNKYp8fjxY7i5ucHKygpKpRJmZmbIzMxEQkICmUZMTAxiYmKQkpIiN4QsLS1FZmYmmQavJrPfffedfJ47OzvjxIkTCAoKwp49e4ymDE6CxYjs8mRmZiI2NlYuWWIxFYdVzwKJZcuWoX///vD398fp06dx7tw5zJkzh1SDB1J2iFarRVJSEuzs7EjW8fPPP6N3797MGzeyhtc1BCjrCyT1FXvw4AFJFpOxIUwETlA0QXR1dWXi+EtcunQJFy9eRFJSEiIjIwGU3SRQN+rRarVM5gwDZen59vb2+Pzzz2FhYYHi4mL89NNP2LFjB0aNGkWmQ93/QJfyKd/29vbYsWMHqS6v4yQRFBSE6OhodOvWDXl5eSQj7AAgPz8fLVq0QOPGjWFlZUUS81nMnTsXN2/exA8//ICsrCz4+/ujQ4cO5F2rV69ejY8++og0psS7776LiIgIPHr0CBEREXo1nIby5Zdfyt/PtWvXwtvbG/b29lizZk2Fjf+LotVqYW1trfeatbU1We0mz0ZJ0k2bq6srevToQZ6NVdm1gsVoUo1Go1dXzqJR7syZM5GRkQE3NzekpKTAyckJlpaW6NevH+kTP5bnHks2btwoTzlSq9UYPHgwtmzZQtpHx8XFBUqlEqmpqfLceDMzM9LJSryazDZv3lzv54CAANL4PJDKYnhsXD766CMcPHgQe/fuRe3atfHhhx+Sa7DqWSAxe/ZsfP311/jPf/6DDh06GPxQ71mwvoaUf2hB1VT40qVL6N27N/PGjbqwOFY8G1UPHToUWVlZSE9PR9++fZlPMKmKCBPBiJg/fz6ys7Nhb2+PnJwcODg4oFq1anj77bfRoEEDg+M3bdoU3t7eqFWrljxL3szMjPzEsLGxwbZt2/R6IVBdqOLj4/Vuni0sLDBgwADyp4gJCQk4ePAgsrKy5Jtmqg1+ZXGys7PlOjIKeB0niTZt2iAsLAxKpVJOy6OgX79+uHHjBrZs2SKP9WnevDl8fX3JNCSkka05OTn47bffcOfOHdy7dw81atQgNV6ysrLIYpXntddeY/b0xdzcHObm5igsLERiYqLcv+PcuXNkGtbW1khJSdFr0JicnFzBWDCUxMREnD59Gvn5+fL5TZ1xIt20AcCePXvIjcmcnBysWLGiQt8W6oadTZo0wcaNG5GZmYmtW7fKG0xK3N3dERoaCisrKxQVFWH9+vWYNGkSQkJCSE0Elufe8zC0JEqtVuPJkydwc3OTN5ejRo3Sm4RkKHl5eWjcuDFKS0vlp30ajQZxcXFkWV+8msxKhl1RURFu3ryJgoICJuYXwG5TeebMGfTv31+vMaEE9bWqdu3aGDFiBAoKCkjj6sKqZ4FEVFQU8vPz8c477+DkyZOIiYmpYCZRwPoaolvCmZmZKU+3MJTU1NQKI2e1Wi0SEhKYZeewOFY8G1XzuE+o6ggTwYjw8PDAxx9/jJo1a0KlUmHPnj0YPHgwVqxYgaVLlxoc38bGBjY2Nhg6dChu3bql152VctxVixYtyGKVR6PRQK1W6zWqU6vV5DXTmzZtwrRp07BhwwZ88MEHiIqKIo0P/F+t6U8//YRr166hRo0aZF2weR0nCaVSiQYNGuDkyZNYsWIFWdyaNWuiW7dusjGRkJAg11a2b9++0tq4F2XVqlVQq9UICAjAhx9+KI/D3LhxI5kGK6Q0/fIjVgG6UVTSONjbt2/r3QBK5RMUjB49Ghs2bICjoyMcHBygUqmQnZ0tjxikYu3atRg7diyz2ngJaext+T9TlJmMGjUKarW6golA0ZtCQqvVom3btlAoFGjWrBnc3d2ZNJFLTk5GQUEBrKysUFBQgCdPnsDKyorcPGINq5KotLS0SjeTFLElEhMT4e3tLY9DlKAsHePZZBYAwsLC0LJly2f2paGA1aayf//+AMo+X2niTo0aNZj00tmwYQPS09P1jhP1Zkn6HS71LKBGq9XK2Qdt27bFrl27mJgIrJEmiCgUCtjY2OCzzz4jiWtra8vEAOYNz2sIr/uEqowwERjw4MEDPWeqcePGGD9+vMFx4+Li5NGLdnZ2SEhIQM2aNcnTuFetWsWsOytQ9oTMz88Pvr6+5NMNBg4ciLCwMHTt2lUepXXmzBkMHDiQVMfCwgKurq7QaDRwdXWtcGNFgdT4Jy4uDnPmzCGtY+Z1nHTp3bs3WrRowaTLsO4TJXt7e/j7+5MaCFqtFp6ennj77bcr/B2r0hxKdNP01Wo1zpw5g3PnzpH0KZCYNGkSfv75Z1hZWckTGZ48eUKaDsmry72npyeT61N5dD8X3T9TZCU8q1Za6uAuPc03BN3u8Cym7EiMGzcOGzZsQGFhIWxsbDB27FhoNBr5e2YssCqJqlOnDvMnYNImz8PDA35+fkxunHk2mQXKNk4sf+fx4OTJk4iOjoazs7N8ryZlklKRmppKXk8usXXrVowePVrP6AbKri2UmZG9evVCcXGx3M/BWCc6NW3aFA0bNpSP1Z9//kkybtPW1lavpI46+5UXPK8hvO4TqjKv7soZ8aymgRSzewcNGoRFixbBzMwMpaWlaN26NTQaDXr06GFwbF1YdmcFym5ib926hUOHDqGkpAR16tQhG9Hm7+8PX19f/PHHH0hMTISjoyOmT58Oe3t7AEBhYaH8BNlQHbVajcDAQMyYMYNJ52hLS0usW7cO3t7e0Gq1pBdBXsdJF3t7ezk+NWFhYfD395dvbCnTeKV4lE3KXgZZWVk4evQo7t69i8DAQMyZM4d0EkCtWrUqnMcuLi5k/S/Ka7Hscp+YmIjJkyfLZRPUN7QA274q/4TU1FSSOCy7w0vUr18fs2bNqvA6dR8J1rAqiaqsBpgVT58+xZ49e6BSqWBnZ4f69eujV69e3PQpUSgUWLFiBerWrSt/d42tsWJUVBRCQ0PlyQnz5s0jNxGcnZ1x5MgRveNElX0iNZJl3Y9m//79uH79OpKTk+Hi4gILCwtmxghL9uzZg/nz5wMo+/7u3buXZB3lm12vWrXqpf+OqurwuE+o6ggTgRiWTQP9/f2hVCpx5coVZGVlyd3VO3bsSKrDujtr9erVYWtri+rVq6OkpIR8pFn16tWfmTmxfPlykguj1EzqjTfewBtvvGFwvMr47LPPkJGRARcXF5SUlJCnbPM4TrywtbWtNEuAktzcXEyfPh316tWDQqGAQqEgKy/RhUX9bGRkJB48eIAePXqgW7duUCgU8nnNYpNv7FCW3PwdUVFRemPgyv9c1WHZHb78U0oAckkOi5s11k0VWZVEjRgxguLt/SMCAgJQo0YN3L17Fw8ePMC9e/eM1kQw1qfRuvj5+cmTExITE+Hn5yeXJVJl/Tk5OSE/P1/OuKQsYdFoNDh69CiuXr0qT4pq3bo1evToQZq1eOXKFSxevBgLFizAggULyBoSlof1NaSoqAglJSUwNzdHcXExWblgeUOWRcZoeYyxia0uPO8TqirCRCCGRdPA3377Db///juKiorg7++PlJQU2YlkAevurNJIlF69eqFRo0bkT7x5EBUVhejoaL0LLfWmOzk5GTt37pRTeI3tCQlPeDxR4tUwh0WHX6VSCS8vL9y9exd3797V+7tXrRHQ89izZ49eFoulpSXq1q2Lli1bMtP09vZ+7s9VHepRqrrwekopwbK7trGXRElMmDABTZo0Qb9+/TBs2DCmWqyaEt6+fVvvZ0tLS7i5ueH/sXfmYVHW6/9/zwwgsoMiBsimoKEommLuuJt7ZYt1ykytk6WnXLJzXBDJPc2w3ErlaOWuqBw1RVlcEHCXXFhnRkQ2YdhxGGd+f/B7njPDoqfm/nwMv/O6rq7Lmbm6H+bzPM9n7ude3re1tTX5sVg/KGVmZiIzM9PgPUFEjson0c9SK5VKJCUlkdgFavdcmUyGefPmwczMDBqNBkeOHCEbtym0uAkl582aNcOtW7eQnZ1ttO2GYK3QP378eCxcuBDOzs54+PAh6XQUfXhUafCYZsByWsbDhw9x6NAh5OXlwcXFBePHjydpLWlKmIIIxLAQDdy9ezdefPFFjBo1Cj4+PkhOTiY/BgDExcU1+P7NmzdJNRE2bdqEgoICpKSkIDQ0FFqtFqtWrSKzz4OTJ08iLCyMaS/UTz/9hBkzZsDZ2Rn5+flYv349E4VZ1mRnZ6Nly5awtLREUlISbty4gf79+5NMFBHgkVGysbHBqVOnkJubCxcXF2YVKCzgGSiIiIjABx98gPj4eERFRaFz587429/+xuXYxqrcd+zY0eC1Wq3GnTt3kJyczOxBz8vLy+C1/rhEY9HpdNi5cyfef//9ep/VLV/9s7CaWw7wy1LygFdLlFarRUpKCqqrq9G5c2dy8cnt27cjPT0dqampiIyMRLNmzTB37lzSYwiwEiX8/fffDV7X1NQgKysLL7/8MgYPHkx6LNYPSvqBAlZ97KmpqUhKSsLNmzfFkcpU3L171yApZmZmhtdee40scCi0uH3wwQeoqanBe++9h5MnT5K10PJEp9OhoqICy5YtQ2lpKezs7MhbN/fv34+rV68aiBM2parUurCclrFhwwa88cYbaNu2LTIyMvDDDz806bX6M5iCCMQEBwcjPT1dHENG0SsfHh4OhUKBxMRE7Nu3D/fv30dKSgrat29P2grw73//G61bt0aPHj3g5OTEbOTRzp07kZ2dDXNzcwQFBaF9+/ZMjsMCoUwwICAA9+/fR5s2bcTPqJ3ax48fo0WLFgBqywlZTU5gzdatWzF//nwUFxfj+PHjePvtt7Fz507SgMiFCxfQoUMHtG/f3ugHycYIDw9Hnz590L17d2RkZCA8PLzBPu0/Q2ZmJlJTU1FRUQFra2v4+voyGVPJA6VSCQC4du0aVq1axWSkJCuV+4ac48DAQCal8+fOnUNUVBRkMhl69OiBcePGQSKRYNmyZWSOiEQigVqtFkUo9aF6EGA1txxgn6UE+N57rFqi9Keu7Nq1S6zIWrlyJblTGxMTg7t376K8vBzOzs5N6vdboKEAmlarxeLFi0mCCDyvKZZTnH766Sfk5+ejQ4cOGDRoEJRKJXnrTGMPwVQPx+Xl5eJYxLt37wKAOFGGCl7nWyKR4MqVKwgODmYmenjjxg0sXbqUiW3g+fJ11Gq1KE7doUMHaDSaZ/wX8ccURCAmIiICjx49go+PD6KjoxEfH48PP/zQaLuenp7i6Ky8vDwkJibiwIEDpA7Cjz/+iN9//x1JSUnIyMiAn58fevXqRf5QNnDgQLi7u5Pa1EelUuHGjRtQqVRwcHBA586dxWyAsRkT/QfftLQ0g8+onbXhw4djwYIFaNmyJQoLCzF8+HBS+yzXSSA2NhaFhYU4d+4cUlJS4OrqigcPHgCorXzx8vIiGQk3duxY3L17F0eOHEFOTg6sra0xa9Yso+3qU1lZib59+wKonVsfHR1NYjciIgIajQYBAQFwc3NDVVUVYmJiEBcXR7J38Ean02H//v2i2BALVWRWKvcNUVlZySSAd+LECSxduhQymQwnT57E6tWr8emnn5IfJysrCyEhIbC3txcfXCmDIvpzy4uKisjmlgPss5S87z1WFUEbN25EYGAgevfuzSz4L+Ds7Iy+ffsyKf1/llAlAXhfUyynOGk0GkilUpibm8Pc3Jw86w0AGRkZ9XwnnU6He/fukdivrKzEnTt36t0XVLoOvM83a20modpZvzWUSjfpefN1unfvjmXLlsHDwwMKhQIvvfTSs/6TuGMKIhAjl8uxePFiALWieyxKW1xcXDB27FiMHTuW1K5MJkPnzp3h7++PmzdvYvfu3SgrKyObK67T6RAdHY2kpCSxNLVHjx4YMmQI2Q/4jRs3cODAAQwcOBCenp4oKirCunXr8PrrryMgIABWVlZG2RfO540bNwxmDLMY8di/f3/07dtXLFujrHRgvU4CwcHByM3NxZkzZ6DRaLBgwQJYWlri0qVLpC0ySqUSSqUSjx49gouLC5PItpeXFzZt2gQfHx9kZGSQjbTLysqq91AXFBREunc01qoEgPQ8AMDs2bORlZWFjh07QqPRYOLEiaT2AXYq9/oCfvrHYiFcpz9xZdiwYfD29sbKlSvFEWRUsMwqAezmlgs2/8j7fxQe954+rFqipk+fjitXrmDDhg3o1q0bLC0toVarmbQZdOvWjdwmb3bv3m1wDdXU1EAul6Nfv35G2+Z9TbGc4vT3v/8dWq0Wt2/fRlRUFAoLC7F37174+/uTaaFEREQ88XNjx/a2atUKEyZM+NP//9Pgfb5ZtyZWVlYiOTnZoG2a6pi814o148aNw4ABA1BQUIDRo0dz0Xj4q2EKIhBjaWmJs2fPwtvbG5mZmeT9iCxJTEzElStXoFar0alTJ8yfP590JN/OnTthZ2eHL7/8Eubm5qipqcF//vMf7Ny50yjHX5/Dhw/jyy+/NMiSBAUFYe3atQgICCA5BgAcOnTIIIgQFRUlljUZy/r16xt1kqkizrzWCQDefvttlJeXo3nz5pDJZNBqteSTJg4fPgxbW1t069YN7du3N2gzoWLy5MnIyMhAXl4ehg4dSjbW08fHB1u2bEHnzp3RvHlzVFVV4ebNm6QCe0IW5vr162jevDl8fHyQlZWFyspK8iDC8ePHIZFIxNJRAEY5gQ3BSuWel4AfUCu4W1hYKAox+fr64vPPP8f+/ftJj3P9+nUcOXIEKpUKq1evRkREBGnWp0ePHvDy8kJJSQliY2Px6NEjMtuss5Q87j19WLVEVVdXo6amBr1790ZlZSUuXLiA3r17kwWDnxWsBNH0f7sBwNzcHOPHj4eNjQ0A40Yc876mWE9xkkql6Nixo6gXk56ejqSkJKaCqvoYO7a3R48ehH9NfXid76tXr6Jr1664efNmvc8oxc+nT58OrVYr6mtQJq943xsAmz3kyJEj8PX1xYsvvggHBwc4ODjg9u3biI+PJ0/u/tUxBRGImTlzJqKjo3Hnzh24uLg0qREma9euhYeHB5o3b464uDjEx8eTjtMSymoFhB9u6n7jumWWlGWXMTExiImJgVKpREhIiLg+lJlv1orXAizXqS6CcwbUOiX6rykICwuDRqNBQkIC1q9fj8LCQmzbto30GPrZ/Pv374vZRGNntE+aNAlZWVlITU3FgwcPYG1tjSFDhpD+sAYHBwMAEhISDPakZcuWkR1DQHA2dTodlEol7t+/T2r/eVG5b6j32snJSfwOGzduxCeffGL0cfbv34+QkBAsXboUUqmU7AFcYMeOHVi0aBH27NkDf39/bNiwgaz6gXWWkse9pw+rlqjVq1dj0KBBKC8vh0KhwCeffILz589jw4YNTXr6CqvM3tP0QIwZccz7mvr9998NgoTHjx9nWhrerl07sgA6D0aOHMnUPq/zLSSWWAvKnjt3DsePH0erVq2Qn5+P4cOHo3///iS2ed8bAJs95NKlS/WCBS+++CJ27dplCiKY+HMIwlVlZWXo2bOn+H5ZWVmTyQbs2bPniZ8b67BptVqo1WoD1Ve1Wk3ab/zw4UOcOXOm3vtFRUUk9gcOHIiBAwfi0qVL6N69O4nNuggaFA2VoFNljVmvE2/WrVuHqqoqODs7Y8yYMUzEvq5evQobGxtxHndJSQns7Oxw6tQpox11nU4HnU6Hx48fQ6vVMhPRlMlkOHToELy9vSGXy5noFeg76B07dsTXX39Nap+Xyv2zJj8/n8SOVCoVHU8W/fKPHj1CTU0NNBoN+vbti1OnTpEfozGMzVIC/O49gF1LlIBUKoVWq4VEIkHfvn3Ru3dvMttCsL+qqgoFBQVwdXVFTk4OWrVqheXLl5Md53kQXuN5TdUNErIaXWiicXicbzs7O8jlcjEhANS2T1Mf68SJE1iyZAlkMhk0Gg1CQkLIgggAn7VivYc0FshpahODKDAFEYiIj4/HuHHjcODAgXqfNeVMgD7GOmyvvvoqli5digEDBsDR0RFFRUWIj4/Hq6++SvY3NjYzd9y4cWTHAMAsgKCPvsOvVCpRWlpKFkTgtU68+Pjjj/90+en/SlVVFT7//HPx9bJlyzBr1iyj5ynzFBuaNWsWkpOTIZfL4eLiwmQ0ptCOo9PpoFKpmMxNZi0u9Twxbtw4hIWFITs7G0uXLiXPlPTr1w+rVq3Cm2++CbVabTAt468Ob6EvVi1Rc+bMwY0bN2BtbW3wd1M6tcJvf3h4OBYvXgxLS0tUV1dj06ZNZMd4HoTXeH+HukFCFoHC6upq8YGMd4uuscLe+sKvdWlKwoo///wzvvzyS4P3WrduTT6BRSKRoLi4GC1btoRKpSLdQ3isFY9j+Pn54dChQxgzZozB1CDKseVNBVMQgQjh4atTp04GUbukpKRn9Sf95QgMDETbtm1x5coVKBQKODo6Yvbs2aLugjF9iALBwcF4/PgxUlNToVKp4OjoiHbt2sHMrOld6voRZwBYsWIFqW2e67R9+3ZMnjy50dfGkpGRgT179oiZuDfeeKNe36uxyGQyHD58GJ6enlAoFDAzM4NWqzXaqeIpNvT48WOYm5vD0dERarUa58+fJ9dEENpxBKE9FpVYrAOz27ZtM3A4qK9XnnTr1g1du3ZFWVkZbG1tyRXWR4wYgf79+6OyshKlpaVkoxd5wFvoi1VLVPPmzdGyZUukpaUhIyODaQb//v37qKyshKWlJSorK8VpOxQ8D8JrvL9D3SAhZSIgJSUFBw4cQPPmzcUe9qqqKrz66qvkv6+sxvYKwq91oZrOwOt863S6er6GpaUledBo6tSp2LZtmxg0mjJlCpltHmvF4xhvv/02oqOjsWLFCtTU1MDCwgJBQUGNJueeZ5rek9VfGK1Wi5iYGPTr1w86nQ5arRanT59GUFDQs/7T/jLY2to2+tBiTB+iQGZmJrZv346AgAA4OTlBqVTil19+weTJk+Hj42OUbQGdTocffviBeeZTX0G6uLiY9MeCxzrpU/cBjPqBbPfu3Zg/fz6aN2+OyspKLF26lNzJmTVrFpKSkqBQKODs7IxRo0ZBKpVi/vz5RtnlKTa0dOlSdO3aFY6OjuS2BaRSKQ4dOoS8vDy4uLhg/Pjx5NUIrFTuBepmLFgFEHQ6HXbu3In333+/3mcNzbL/I2zYsKHRzyiDMJs3b0ZhYaHBNcWr+s7YLCVvoS9WLVE8s9/Tpk3Djz/+iMrKSlhZWWHatGlktnmfj/Xr12PGjBniayFYaMxUC97fQT9IaGVlhStXrpDZ3rNnDxYsWIBmzZqJ71VXVzP5fWU1ttfYffRp8DrflpaWyM3NFUcnA0BOTg55ZYinp2e9igcqeKwVj2NIpVIMGzYMw4YNI7PZVDEFEYiIjY1FXFwc5HK5qPBtZmaGrl27PuO/jA5jHTYe/Pzzz5g7d67BVIkRI0Zg3bp1ZJFIiUQCBwcHpKWlwcfHh5ngjf6PtJWVFTw9Pcls81inJ5Genk4qziQIXALs+tLUajXy8/NRXl6O0aNH49q1ayRzgXmKDVlbW5O2DzXEhg0b8MYbb6Bt27bIyMjADz/8QH5NsVK5r6tDYmFhgTZt2sDd3d1o2w0hkUigVqtFTR19nib+9jQE5zkqKgq+vr7w8fGBXC4nH0ebn59vdEvP02CVpeQt9MWqJYpn9rtdu3aYNm0aioqK0LZtW6jVajLbvM5HVlYWsrKykJmZKWoDabVaUWvFmOopXt8hPT0dR44cAQC8/vrruHv3LuLj49G5c2eyxJW5uTmUSqVBpYxSqYS5uTmJfX1Yje0ViI+PR3R0NLKzs2FlZQUbGxuS6k5e53vy5MnYvHkzHB0d4eDggKKiIpSUlJBN4xDGG+v7Umq1GiqVChs3biQ5Bo+1ehbijf+XMQURiAgODkZwcDD5w9GzgJXDxgOdTldvLKW9vT15yVdGRgYyMjIM3qN22BQKBV555RUAtd/rxIkT4mtj4bVOjYnm7Nq1i/TB46233sKyZcvEH8E333yTzLZAeHg4RowYgUOHDkEmk+HYsWMkQQSAnxCXRCLB6tWr0aZNG9FRoC4/V6vV4rjTDh06QKPRkNoH2Knc173+y8vLERkZCRcXF2YZLWFqjb29vajvQDGxRgj6yuVysZLC1dUVv/32m9G26x7n2LFjBtcU9fg3VllKgK8IHquWKJ7Z74MHD+LevXvIycnBihUrsGbNGvzrX/8is8/jfEgkEnFKkEwmg06ng4WFBZmPw+M7/Pvf/8YXX3yB8vJyhIaG4p133sGSJUtIg+gzZsxAZGQkfv31V2i1WkilUnh6ejKpwmQ1tlfgPAWjLQAAIABJREFU+PHjWLp0KZYsWYKFCxeSannwON+tWrXCwoULUVhYKAad9Sv8jBU+1x9vXFRUhN9++w2pqankv3s81ornnv5/HVMQgRgzMzOEh4dDpVKJDmlT6ucD2DpsrHF2dhZnZAucP3+evIoiJCQEGo0GpaWlcHJyIrUtkJSUJAYNJBKJwWtj4bVO77//foP9vgqFgvQ4nTt3Ji+vrEtNTQ0CAwNx+PBhAHRK9zxLkVkIKdblpZdewrJly+Dh4QGFQkEWaNGHlcp9XR0SABg6dCgWL17MLIhANQ6xMfz9/bF8+XJ4eHhAqVQa5Wg2RIsWLVBRUSFWOFD1GuvDKkvJWwSPVUuUfvYtJycHNjY2zLJvN2/eREhICEJDQyGRSEiDhLzOh5eXF7y8vBAcHAylUomKigrodDo8fPgQLi4uRtnm9R3Mzc3h5OQEJycnuLq6Njgy1lgcHR25aMHwGNvbrFkzSKVSyGQyFBcXk0344b2HtGzZssH2QIpJNZmZmTh27BgePXqE4cOHY+LEiUbZq8vzIqwocPz4cWaJvqaCKYhAzI8//ogZM2Zg8+bN+PjjjxETE/Os/6Q/DMuyMpVKhRs3bkClUsHBwQGdO3cWgxPG9CEKTJ06Ffv27cOJEyfErLSfnx+mTp1qtG19YmNjcfbsWZSWlmLlypVYt24dZs2aRXoMAHjw4AFeeOEFPHjwgLRKgNc6ubm5Ye7cufXKQ8PCwkiPI/xI6IvGUQfvXnjhBURGRqK8vBxRUVFkJe48S5H9/f2Rm5tLrrGhz/jx4xEcHIyCggKMHj2aSfCRlcp9Q2g0Gjx+/JiZ/evXrxvMeY+IiCB1dt566y2oVCoUFhZi1KhRZOdDyO68/vrrJPaeBKssJW8RPFYtUQDg5OSEli1bonnz5rC2tmame2JmZiaOAlapVKSl7bzPx4oVK+Dk5CSulUQiMbqNiNd3yMzMREhICHQ6He7duyf+m6qSCahtXdi9ezeqqqpE21ZWVnjzzTdJ2yt5jO0dP3481Go1JkyYgK1bt2LkyJEkdp8HMVCg1l8qLy/HiBEjRN0FYbIFVVD4eRFWFGCZ6GsqmIIIxJibm6N169bQarVo3bo1ef8pD1g5bDdu3MCBAwcwcOBAeHp6oqioCOvWrcPrr7+OgIAAEhV3CwsLvPvuu0bbeRpnzpzBkiVLEBoaCqlUirKyMvJjTJs2Db/88gvKyspgZ2dHFpEH+K3TV199BQsLi3rvC+WvFBM5AODkyZMICwtjOoVj6tSpuHz5Mvr164fWrVuTZfV5liJv3boVlZWVuHv3Lvz8/KBWq412musSGhoKf39/vPzyy8yql1ip3AvjKQXUajUePnzIVEei7pz3e/fukdpXKBSIi4sTs60ATWvakwKBlA4byywlbxE8Vi1Rhw8fxu3bt9GpUydYWVnhwYMHiI6ORvv27ckVw6dMmYIdO3agoqKCPODF+3zodDrS31WA33eIiIh44ufGlrcDwE8//YTPP//coNpS8Nv0y98pYDW2Vwh+CJWK7du3J0lYCfC+ZlkhBApUKhVUKpX4PmVl2fMirKgPq0RfU8EURCAmMDAQarUaffv2xdy5c412annD0mE7fPgwvvzyS1hbW4vvBQUFYe3atQgICDDKtsCTIvCUjq1MJkN1dTWA2gcNFmJ+rq6umDNnDnQ6HTIyMkgV7nmtU2PZMJlMBoBmIgcABAQE4P79+2jTpo34HvU5EZS8Baefauxf3VJklkJASqUSoaGhCA0NxcyZM/HNN9+QH2PhwoW4c+cOTp06hfv378PPz49cd4GVyr0wnlLA3NzcIBAiOAyU1J3zTs0PP/yAKVOmkLdd1b1vheoyalhmKXneewC7lqgrV67U29NHjx6NkJAQ8iBCWlqagThkUlKS0S0AArzPh5WVFSIiIgzaoQYNGmSUTd7foTEoytt5wkpza8eOHZg0aVKDQU+K9fmrnG9jW1FZT7EA+KwVz/PBMtHXVDAFEQjR6XRwcHCAhYUFhg4dSj52jAesy8r0AwgNvTYWd3d3FBcXo127dggKCoKrqyupfYF3330Xa9aswf3797FmzRry3jGgVuhm0aJF2LNnD0pLS/Hw4UP885//JLHNa514kZaWhrS0NIP3qByopyl5U+Dt7c3F6ZBKpdBqtbCyskJcXBxyc3OZHMPX1xfl5eUoLy9HSkoKeRCBlcr90xyxLVu2kDvmdee8jx07ltS+h4cH2rZty7RKBwC+++47Zg8trLKUAL97D2DXEuXs7IzIyEgEBAQYZN9atGhBYl9Aq9XizJkz6Nu3L7Mx1jzPR5cuXZjY5fkdWDJt2jRs27YN5eXlYkbfxsaGvO0RYDe2V2jBffXVVw10kyirhHme76YsfA7wWSte54Nloq+pYAoiECKRSHDlypUGxbmaEqwctocPH4oPYfoI/ZUUTJkyBUDtQ2VsbCyys7PRs2dPDBgwgOwYQO2YK2PFsJ6GkKUqKCjAjBkzSKcZ8FonXsybN89A4Vy/HM9Y9JW8hYwxpZI3T2bOnAmdTodp06bh3LlzTFS2v/vuO5SXl6NLly545513mPywslK5fxboz3m3tbU1aKegQKFQYPr06WKfK2XPtD6sRqsCTcdBfhqsWqKmT5+OS5cuISEhARUVFbC2toafnx/GjBlDYh/gN8Y6MzMTqamp4vfw9fVF27ZtSY8hQC0A+rzRpk0bzJkzh8uxWI3tFTh06JBBECEqKkqcItSUaMrC57zgtYewTPQ1FUxBBGJYZkx4wcpha6yscty4caTHqa6uRkFBAYqLi2FhYVFvlCEFMTExiI2NNXCcqbNwLVq0QFhYGAYMGIDHjx+TO+k81okXQquBpaUl8vLysGXLFrKgi76Sd01NDUpKSkjsPomSkhLs3r0bffv2RceOHcnsCu0lDg4OTCY16HQ6uLm5YcKECeS29WGlcs+TDRs2NPoZ5R68evVqMlsNkZSUhG7dupEGOevCKkvZEKzuPYBdS5RUKkVQUFC9igDKyQk8xljzVrrfu3cvgNp9Kzs7GzY2Nk1qD3kSFJOWioqKcOjQIdy/fx9arRYSiQTu7u4YP348eZULq7G9MTExiImJgVKpNBCfZBWYAtjuISyFz4Haykv9LH7d19RQrxXPPYRloq+pYAoiEPM8ZExYOWzBwcF4/PgxUlNToVKp4OjoiHbt2pGW2a5cuRIVFRXo2rUrRo0aJYr25eXlkfVtAnyE/D777DM8fvxYnGM9b948Mtu81klAcJ4FBOeZSuDo7bffRnh4OEaPHo1Dhw4ZHIuKyMhIXL9+HTk5OXBxcYG5uTmzHw1bW1uMGzeOfBQmayQSCZRKJfPjsFS5fxINiYT+WYQe1KioKPj6+sLHxwdyuZxcjFcul2Pfvn2oqqrCggULEBkZ2aDmzZ+lsLAQK1euhKOjI/r06YOAgADygCfrLKU+LO49Hi1RDbFixQosWLCA1CbLMda8le7r+mvffvstk+OwfKhkWd7+/fff45133jEIGqWnp+OHH37AokWLjLavD6uxvQMHDkRwcDAuX76M7t27k9h8Gix/v1kJnwsoFAqDoEHd19RQrxXPPYR1oq8pYAoiEHPr1q167zW1EnFWDltmZia2b9+OgIAAODk5QalU4pdffsHkyZPh4+ND8JfXbki2trZ48OABHjx4YPAZxY+qMNaMh5BfQyXHVBsh63USeJrzbOxEDmEEEVBbmhoeHo7JkycjOzubvLzv0qVL+Prrr7F48WIsXryYmcMJ1F5L5eXl6NmzJ7NjsIJHNRYrlXug9qH70qVLYqDzpZdegpeXFwCQlioKmUK5XC5mo11dXfHbb7+RHQOoDdjNnTsXa9asgVQqRUpKCmkQYeTIkRg5ciQKCgqwc+dOfP/99+jVqxdGjhwptlAYC6sspT5KpRL37t2Di4sL2rVrR/a3A+xbovR/F4RMqzD6jxqWY6x5K6vr/34UFRUhLy+PyXFYPlSyLG9Xq9UGPg5Qq6ekVquNtl0XlmN7JRIJzpw5wyyIkJmZiRYtWsDW1haXL1+GhYUFunTpQrqHAGyFzwXqtmNTt2ffu3cPUqkUbm5uACBON6PydXjuISwTfU0FUxCBGH21ZaVSidLS0iYXRGDlsP3888+YO3euQdn8iBEjsG7dOrKH46c5Zbt27TJKBFFf4ZeVkF9D9hQKBRITE8lss14nAX3nWdhoKZ3n27dvG7wePHgwlEol6VgiAaHqpFmzZrh16xays7NJ7AqBqbrs2rWLrNJh0aJFYq+9/h7Foj+eRzUWK5X78+fP4+zZsxgzZgwcHR1RVFSEvXv3ok+fPujTpw/JMeri7++P5cuXw8PDA0ql0uixbA1hY2Mj/vvx48ektrOzs3Hu3Dmkp6ejbdu2ePPNNwHUToV40hjIPwKrLOXSpUsxf/58/Oc//8HNmzfRrVs3HD9+HE5OTqQjcFm3RJWWlmL16tX1KuOo1l8flmOseSvdC78fEokEVlZWzPr/T548iREjRpA/VAJsy9vffvttrFixAs2aNUPz5s1RWVmJmpoacqFcgN3YXn2WL18Ob29v8beQ4nts3LgROp0O5ubmKC0thaOjI6ysrHDx4kV8/PHHRtvXh6Xwub6PIKBWq6FSqbBx40aSY+zYsQMqlQoymQxlZWWYPn067Ozs8Ouvv5L5zzz2EKGStq5fxUpv6K+MKYhATN2o3YoVK57NH2IErBw2nU5Xr+/e3t6e62zV1NRUo/5/YaO7ceMGM6XfhnBzc8O1a9dEB501xq6TgL7zrFQqxVn1Dx8+JGmb0B9LVFlZicrKSqNtNsYHH3yAmpoavPfeezh58iTee+89Ervvv/9+g44SZdaKeqb3k5BIJDh48CCqq6vx2WefISYmBoMHDyY9BiuV++joaPzzn/8U2xZcXV3h5+eH5cuXMwsivPXWW1CpVCgsLMSoUaPIK2gGDBiAVatWIT8/H2vXrsXAgQNJ7Z84cQL9+vWrNx6T8iGcVZZS0AxISkpCSEgIpFIphg0bxqxNiVVL1KRJk6BWq+sFESjPgQDrMdY8le7feOMNpKenIzc3F61btyYRgW0oYHvv3j0kJCQwecBgWd7eqVMndOrUCWq1GuXl5bCxsSFt6dKH1dheARYaQACQm5srntfZs2dj9uzZAJ48RtsYWFX66fsIRUVF+O2335Camko6+jEjI0NcF4VCgbVr1+Jvf/sbmX0B1nuIUDnI06/6q2IKIhCzZ88e8d/FxcVcH5CpYOWwOTs748KFC+jdu7f43vnz50kEgHjDQ+lXcEaEa4j6QYwnK1asgJOTkyjsJ5FI4O/vT2Z/8+bNKCwsFO0DdBnxzMxMSKVSsaTd3d0dgwYNarSC4I/i5uaGuXPn1mvtYJFFrK6uxoULFwwmolCLIG7cuBEffvghfvrpJ0ilUpw/f5782mWlci+VSus5yRYWFkx7HRUKBeLi4sQAG0BbzTFo0CD06NFD1DuxtbUlsSuUgvfs2RM1NTX1Woso90NWWcrs7Gx8//33yMvLg0ajEc99TU2NUXYbg1VLVGBgYIPvU7UJCjwPY6z1iYiIwKNHj+Dj44Po6GjEx8cbLb4WFBQEhUKB4OBgUf9g2bJl+Ne//kXxJxvAo7wdqN0DnZycANTqO+zZswd9+vQh1XdgNbZXwN/fH7m5ueR+ub4foF+9ycr3Z1npl5mZiWPHjuHRo0cYPnw4+ehyrVYLjUYDMzMzeHp6Ys6cOVi/fj1ZVScv9u/f3+hnrEWl/2qYggiEyOVyyGQynD9/Hn369MGIESPg6en5rP+sPwwrh23q1KnYt28fTpw4IT4c+/n5MZk5zAqeSr/PU5RTp9OROjV1yc/PZ5Y9/Pnnn/Hll18avNe6dWusXLmSpATvq6++ajC7w8LpXLNmDXr16oXk5GQMGzYMGRkZ5MfQarVivyPAxplipXKfm5trEAgGav/+3Nxco203xg8//IApU6aITjoVqampsLCwgJeXFw4fPozS0lIAtZUJFM5/3VYiARatRKyylMuWLQNQWw0iBIqqq6uZlGsD7FqiePEsxlizFCWUy+VYvHgxAGDo0KEk+/no0aOh0Whw+vRpnDp1SmwNZQHL8vbGsLW1xdixY8n1HViP7d26dSsqKytx9+5d+Pn5Qa1WkyQyPvroI2i1WkilUlFzQaPRMKt8YCV8HhISgvLycoO2GyEwTLWfT5o0CRUVFbC3twdQ+13mzZuHhIQEEvuNQb2HJCQkwNLSEj169ICfnx/BX9h0MQURiBB6aceOHYvevXuLvbR9+/Y1yLw3BVg5bBYWFkzKK/8Ixm64AwcOxMCBA3Hp0iVmIj1PKoNjpVJdF+oMk5WVFSIiIgxaYwYNGkRm39nZGceOHUObNm3EUlKqHz6dTlfPkbG0tCR7ONavntBHJpOR2NdHo9Fg0KBBiIuLw+DBg5GUlER+jI4dO2LLli0oLi4WhVSpYK1yP3ny5AbFPvWrjqjx8PBA27ZtySe97NmzR5x+kpqaipkzZ0Kj0WDr1q0kjtTrr79utI3/FVZZyoaq4CwtLdG1a1ej7DYGq5YonvAeY81SlNDS0hJnz56Ft7c3MjMzyR5YzczMMHz4cAwZMgTx8fFMk0msz0dJSQkyMjJQXl4Oa2trtG3bFq1btybXd2A9tlepVCI0NBShoaGYOXMmvvnmGxK7dYUngdrzz8o/ZCV8LvhLKpUKKpVKfJ8yKNxQVbNUKmXWKihAvYesWbMGDx48QGJiIo4dOwYnJycMGDCA6djQvyqmIAIRT+qlbWpBBFYOG8+H4/379+Pq1asGGd6QkBCyc+Hs7IwVK1agqqoKVlZWeOutt8Ryd2Nxd3dHcXEx2rVrh6CgILi6upLYbQjW6yTQpUsXUnt1adGiBSoqKkRtCsofPktLS7FnViAnJ4fM4eSJnZ0d1Go1PDw8sGHDBlRVVZEfY8KECVAqlQgICICrqyupA81a5f4///kPt0CdgEKhwPTp08Xri0qcST+LN2zYMLHfm6oN50ntNtRryDpLyRrWLVECPGa8sxZOzczMRGpqKioqKmBtbQ1fX1/xwZWamTNnIjo6Gnfu3IGLiwv+8Y9/kNqXyWTkGiR1YXk+Dh8+jNu3b6NTp06wsrLCgwcPEB0djfbt22P8+PGkx2I9tlcqlUKr1cLKygpxcXFMq8tYwkr4/GnaBxs3bsQnn3xCcixesJy488ILLyA4OBjW1tY4f/48bt26ZQoimPjzPIteWlawcth4PhzfuHEDS5cuZWZ/69atmDFjBpydnZGfn4/169eT9bBPmTIFQO30h9jYWGRnZ6Nnz55MpnywXicB6vLmulCK/9Rl8uTJ2Lx5MxwdHeHg4ICioiKUlJTg73//O7NjsmLq1KmwsLDA5MmTIZfLxbJCCoTKAH0yMjKQkZFBVnXCWuUeqK08aajKhNVevnr1aiZ2JRIJqqurYWlpKTqdlZWVZA+uPIMtrLOUrGHdEnX8+HG88sorXGa81x1jLZPJoFKpSAQWIyIioNFoEBAQADc3N1RVVSEmJgZxcXFGaxU0hJWVFQYPHozy8nIAtVl9Y8cO84ZVeTsAXLlypV5Ac/To0QgJCSEPIrAc26vVavGPf/wDWq0W06ZNw7lz55hWz7CElfD508jPz+dyHGPhMXEnKioKqampcHBwQPfu3bFo0aIm+axHgSmIQMSz6KVlBSuHjefDsY+PD5KTkw3K2ymmAQg8fvwYLVq0AFCbBafOKFVXV6OgoADFxcWwsLCoN9WCCtbrJLB3714AtfdEdnY2bGxsSB4AGsrYSqVSTJw4kUwQtFWrVli4cCEKCwtRXFwMR0dHAxXv27dvMxnLx4K1a9eKKvQ+Pj5Yu3YtZs2aRWJb/0dUIpGgoqIC0dHRUKvVpK0rADuVe7lc3qgWCauHZrlcjn379qGqqgoLFixAZGRkg0Jpf5QJEyZg5cqV6Nevnziu8ty5c+TCTw3dg9RrxTpLyRrWLVFJSUl45ZVXmM94B9iq6GdlZdW7noKCgpjdeywFeXnBqrwdqK24jIyMREBAAJo3b46qqircvHlT9H0oYTW29+TJk4iNjUXr1q2Rl5eHV199lZleAQ9YCZ8/L/CYuHPixAk4OjpCpVJBLpfjwIEDphGPJozj008/bfB9lr20rGDpsPF6OK6srERycjKSk5PF9yidgxEjRmDBggVwdnZGQUEBRowYQWZ75cqVqKioQNeuXTFq1ChxZJOgrk4J63VqzCaVKnlDzmV5eTlWr15Nvpm3bNmywRFge/fuJXFyDx06hFdffbXR18aQlJSExMREUY0eqA2EUbYzCA8sDx48wLFjx5CXl4c333wTQUFBZMcQYKVy7+Xlxb2dYfv27Zg7dy7WrFkDqVSKlJQUkiCCv78/Zs2ahatXr0KhUMDR0RFffPEF+Z6rv14KhQKJiYmk9gG2WUqA7b0HsG+Jys/Pr5fEEKAWiWSpou/j44MtW7agc+fOBg+trMa15eTkMHP6WV9TAqzK24Ha3+1Lly4hISFBbC/x8/PDmDFjyI4hwGpsb3x8PL7++mtIpVI8evQIK1euZKJXwOt8sxI+5wnLteIxcUfwoUyYgghkUI6re9awcth4PhxPnz4dWq0WJSUlsLe3Jy816tevH/r06YPS0lLY2dmR2re1tYWtrS0ePHiABw8eGHxG/YDPep0E9Me/FRUVIS8vj8lxABj0yzcl6vbOUvbSdurUCd7e3mjZsiWGDh0KnU4HmUwGBwcHsmOkpKTgt99+g6WlJV555RXy8XL6sFK5f1bjZm1sbMR/P378mMyura0t+vfvT2bvabi5ueHatWt48803Se2yylIKsLz3APYtUdbW1qQCpk+CpT7FpEmTkJWVhdTUVOTk5MDa2hpDhgxhFkTo06cPM0Fe1teUAMvydqlUiqCgoHqB4JqaGnLhX1Zje3U6HQoLC8U949GjR6L/Qel38jrfLCuBgFphRQcHB6jValy+fBn+/v6wt7cnbRlluVa8J+78X8cURDBRD1YOG8+H43PnzuH48eNo1aoV8vPzMXz4cHJnWiqVkj6ECfAsp+SxTsB/x8FJJBJYWVlhzpw5JHb1gxNA7bV79+5dJi0ZLLl3716966mgoIDs+rKysoKVlRWGDx+OxMREVFRUiJ9RlbeHhYXBxcUFDg4OiIiIEMe4sijxY6Vy//LLL6OyslIUEROyu6+99hqz3tMBAwZg1apVyM/Px9q1a5kLsVGzaNEi8VwDwODBg8mPwSpLKbQn2dvbIykpCffv30erVq3w8ssvk9gXYN0SZW1tzS2RwVqfwtvbu17QID09nUnZ9rVr1+Dg4CDuh5SCvA4ODvWE3VjAsrz93LlziIqKgkwmQ48ePTBu3DhIJBIsW7aMvGKL1dheNzc37N+/3+D1gQMHAND6Wvq/1SdOnCCtTtWHZSUQUJtEXLRoEXbt2gU7OzucOHECoaGhZPtLZmam2AJ8+fJlWFhYkApvNzZxx9bWluwYJv6LKYhgoh6sHLanbdi7du3CxIkTSY514sQJLFmyBDKZDBqNBiEhIVwzck0FXuv0xhtvID09XSzpbagt4M9Qd1a9ubk5fH190a1bNxL7/wvGZq937NgBlUoFmUyGsrIyTJ8+HXZ2dvj111/JHbXVq1dj1KhRTEaONVZOLUChHcFa5f7gwYP4+uuvAdSWLH7wwQewtbXF5s2byc9FamoqLCwsMGjQIKSmpsLd3R1qtfqZVUP8URISEtCrVy8sWbIEOTk5TIVyWWUpw8PDERISgoiICFhYWKBTp06Qy+X47rvvyLRC9GHVEsVTWZ1lu2Nj9/GuXbtIe5r1j/fRRx+R2uQh7KYPy/L2EydOYOnSpZDJZDh58iRWr17daOvun4X12F7WSRkhiAr8N+F27949JCQkMGmVYT2pRij7Lysrw6RJk3Dt2jUSu0DtPqTT6WBubo7S0lI4OjrCysoKFy9exMcff0xyDB57SGFhYaOfUfm2TQVTEMFEPVg5bE8jNTWVzJZEIkFxcTFatmwJlUpFXt4eFRWF3r17w8nJidSuPmVlZcyjp6zXSSAiIgKPHj2Cj48PoqOjER8fT6K2zdN5XrJkCTp37ozevXujVatW4vvGOikZGRmis6FQKLB27Vr87W9/M8pmYzg7O4v9s7yh0I5grXIvtEkUFRVBq9Uy7TPds2cP5s6dC6C2N3vmzJnQaDTYunUrOnbsyOy4VJw8eRK9evUCAPz4449MtSRYZSmF/S47O1t0MLt06dLkxLGeliWkVFZnqU/x/vvvN3jPUc13r4uVlRUiIiIMqoyMFYHlIeymD8vydqHtDagdE+vt7Y2VK1eSTsVhPbaXNUFBQVAoFAgODhb37WXLluFf//oXk+OxrgTy9/dHaGgoxo4dC7VaXW/qnDHk5uaKe+vs2bMxe/ZsAE8e//5H4bGHCELhRUVFKC4uhoeHB5RKJRwcHJjd539VTEEEE/Vg5bDxZOrUqdi2bZsoBiRMhqDCxcUFP//8MyorK9G9e3e8/PLLBn3NFISHhzMfYcZ6nQTkcjkWL14MABg6dCg38TpK53nBggW4efMmDhw4AJVKhcDAQPTq1cvolgOtVguNRgMzMzN4enpizpw5WL9+PVmfvz7l5eWYN28e3N3dxexJUxp1xVrl3snJCXv37kVmZqY4NUatVosPBpToZ4+GDRsmZjCoJ700ZVhnKQcMGIBNmzahRYsWCA8Ph7+/P5RKJVM9j6YOS30KNzc3zJ07t96YRarxyXWpK3wt7InGwEPYTR+W5e2DBg1CYWGhuDf5+vri888/N2gPMBYeY3tZMnr0aGg0Gpw+fRqnTp1iHqRnPanG3d3doCKYcmSo/m+b/jGa2h4iBLi++eYbrF69GlKpFFqtFt988w3ZMZoKpiCCCRHWDhtPPD0962UsKenRowd69OiB8vJybNmyBTt37kSXLl0wcuRIdOgmdIDIAAAgAElEQVTQgeQYjo6OiIyMhI+Pjxihp+rXFGC9TgKWlpY4e/YsvL29kZmZSVZ6xxOpVIr27dujtLQUFy5cwO3bt5GWlgZ7e3tMmjTpT9udNGkSKioqYG9vD6BWZG/evHlISEgAUOskCiKkxkJVlfGsYK1y/9lnn+HatWvw9/cX77WKigpRc+HBgwd44YUXSI4lkUhQXV0NS0tL0fGsrKxsMkGE4uJihISEQKfTobS0VPw3pQYG6yxl//790alTJ1y/fh0lJSXQarUYNGiQ2C5Dee89iabSwgKwa3cEgK+++qrBzCerrK4wUUaj0eD69etISkoyeuT004TdqK8pluXtdfVNhMoE6hYQgN3YXgFBi6Sx18ZgZmaG4cOHY8iQIYiPjzdoF6Q+3ywrgbRaLc6cOYO+fftCp9NBq9Xi9OnTZBOWPvroI2i1WkilUnFChkajEaudKdaK5x5SXFyM9PR0eHl5QS6XN7kAGAUSHbXMsYkmi1wuh1wux+nTp8UfDzMzM/j6+nIRqrtw4QJ69+5tlA2hakK/T02AsmTqypUruHDhAioqKvDSSy+hV69e0Ol0pKMF9+3bZ/BaIpGQieDxWieByspKREdHi5M4hgwZUi9SzILQ0FCyqod169ZBrVYjKCgIPXv2FH/stmzZwsSpEqD8DoWFhTh48CCqq6vx2WefISYmhokQXkNs2LDB6Ie//Pz8RlXueTyIUZ6LW7duYd++fejXrx8cHR1RVFSEc+fOYcKECU2inYE3dbOUPHpPKc830HhLFBWNKavfunWLVHjx8uXLuH//PlxdXZmMy+NBVVUVLl++jKtXryItLQ1jx46Fv78/U20PgP6a0mg0SEpKQkFBAZydnREUFCS2ZVHwyy+/YMyYMbCxsUFYWBjMzMzg5OREHpBesGBBvbG9X3zxBZn9w4cPY9y4ceLryMhIjB8/nsx+Y1Cf77CwMCxcuFC0u2TJEixatMhou7GxsYiLi4NcLheDqGZmZujatStGjhxptP3/Beq1Yk1+fj6OHDmC/Px8tGrVCmPGjGlyot7GYqpEMCHCq6xs//79uHr1qkG0MCQkxOgAAgCx7WLq1KkGfY45OTlG29YnOzsbEydORIsWLQzepxKHAWr7/VNTU6FSqdC9e3c8fPiQzDavdRKwsrLC4MGDUV5eDqBW74EyiMBjLFG3bt0MRCfv3LmDDh06MA0gULNx40Z8+OGH+OmnnyCVSnH+/HnyIAIr7QiAvco9T/z9/TFr1ixcvXoVCoUCjo6O+OKLL2BnZ/es/7S/HKyzlLxg1RIlwFpZHeDT7shyjjwALF++HBYWFujZsyemTp2Kb7/9FkOGDCGzzxOW5e2xsbFITk6Gu7s7CgsLIZFI0KdPHxw9ehRxcXHw8vIiE+llNbYXqO1d79ixoziSGwCXAAILWFUCBQcHIzg4mNkkFN6w3kOAWn/knXfeQWVlJandpoQpiGCiHqwdths3bmDp0qVk9hpi+/btBhHN3bt3k6pt37x5E2PHjq33PmUWY+vWrbC0tERKSgqCgoKwefNmLFiwgMw+wH6dBDZv3ozCwkKD8kFK4SQeznNMTIxBECEqKoqsdYUXWq0Wbm5u4msWhWisH5QAdir3vLG1tW2yU2OE1gUeXLp0qV6WsinCqiVKgKWyOs92R5Zz5IHa3v60tDSkpqbCycmJyT7IC5bl7cHBwYiLi4OjoyOuX7+OiRMnwtfXFzdu3DC67aMurMb27t69G3K5HB4eHsjIyECvXr2abMAIYCd8LgQDd+zYwaU6lTWs9xCAvV/bFDAFEUzUg7XD5uPjg+TkZLRp00bcrKhKgGJiYhATEwOlUmnQo1u3YsBYeOgV3L9/H4sWLRI3cMpeaV7rJJCTk8P0h4il89zYWrVt25bsGLzo2LEjtmzZguLiYmzfvh0BAQHkx2D9oPQsoVSqbuqEhoZi8eLF2Lx5M2kFVkOwzFLyRL8laubMmQYtURSwVFbX16eQyWTQ6XTkKvqXLl1CQEBAvYAjZQASgNgWmJ+fj6SkJGi1WmzYsAEvvvgik4cNlrAUugRqH1p/++03dOzYEb6+vqipqYGfnx+ZfdZje2/duoUlS5YAqPWhwsLCmnQQgVUlUGBgIIDah2B9zaGmhkajwbVr12BrawsHBwfEx8ejsrIS/fr1Iz9Wfn5+k6yIo0S2WJBMN2Hi/xMfH4/g4GAkJCSgRYsWiIuLw/Dhw8nsX7hwATk5Obh9+7b4X48ePUhse3t7Y+DAgbCxscGUKVMwcOBADBw4ED179iSxLyCXy6HVavHw4UMUFhbi4cOHpFlvAEhMTESrVq1w8+ZNeHp6Ij09naTlA+C3TgI1NTW4e/cuqqurUVBQIPaQUZGXl4eDBw9i6NChaNGiBRITE8kyvMJaubu749133xXXqkuXLiT2BcLCwgyyO+vWrcPLL7+MoKAgmJubkxzD398fzs7O8PDwQOfOncURfZSsW7cOFy5cgKurK15//XX0798fPXv2RFJSEqmKdEPExcWJYmnGUFRUhNjYWFy/fh23bt0Se8pZOCJNlYsXL+Ly5cu4fv06FAoFkpKSkJiYiOTkZDIhLgEvLy9YW1ujbdu2iIuLw4gRI0gdXR73HlBbev7OO+/Ay8sL5ubmuHPnDlq2bEl2X6hUKrz//vt44YUXIJPJYGlpaVB5ZAwODg7w8vLCoEGDIJVKYWFhgebNm6OqqopMjyQ0NBTJycm4fPkyampq0LJlS6aBO2tra/j5+WHAgAHw8/NDdnY2fHx8kJCQgDZt2hhlm9c1lZaWBqVSiYyMDKjVashkMnTt2pXMvp2dHbp27SpOLJHJZKTl7uvXr8fw4cMNdBwsLS2xYcMGkr382LFjcHV1RX5+PgoKCpCcnCy+pvRBWJ/vrKwsXL16FYmJibCwsEBWVhYyMjJw48YNksBXeHg4goOD8e2332LAgAHQ6XTif9QVZyzXas2aNZDJZFAoFIiMjISjoyOsra2xb98+8qq/27dvIz8/n5lf2xQwVSKYqAersjKB6dOnQ6vVoqSkBPb29mImn4KYmBgMHDgQhYWF2LNnj8FngjoyBZS99o3x97//HZGRkbCwsMDZs2dJe+95rZPAtWvX4ODggIqKCgC1WS3Kyg2WY4mEXrqEhARcvHjR4DOK8YgpKSlISUlBbm6ueC4eP36M4uJiACDRjoiLi6v3niCkSlmWqtPp4OHhgddee63eZzy0I6geZlavXo1Ro0aR9fs+j3z11VcoKirCL7/8grfffptJSTjrLCWPe08fli1RrJXVBVasWAEnJyexhFcikZAF0F1dXRESEoL8/HwkJibim2++gbm5Obp3706ayGgIe3t7MUN98uTJPx1g5X1NsSpv16dutptSB4P12N6goCDcvn273msqH4TX+WY9qWbIkCFYuXIl5HK5WLkhQNUiyGOtKisrRf9j9uzZGDNmDICGfSBjadGiBSoqKnDnzh3xPeqK5L86piCCCRHWDpvAuXPncPz4cbRq1Qr5+fkYPnw4adYYqD//mRr90vycnBw4OTlh+fLlpMews7PDhAkTRNEWtVpNZpvXOglotVpmD5GsnWchyv/222+T2KuLi4sLpFIp8vPzxfYCmUxGKvz073//G61bt0aPHj2Y9gBLJBIuI2FZijcCtcEI1vO+nwecnJzw6aefIjExEbm5uXBxccHLL79MZv/nn3+uN4K2devWWLlyJYljy+PeA9i3ROkrqwsPAIKyOjU6nY55QFBQOh8zZgxUKhUuXbrE9HiU8LqmBHgIXda1R2mf9djepyV8Nm7caNSkCV7nm7Xw+eDBgzF48GBcvHiRdA/Xh8daNWvWDAcOHEBNTQ0cHBxw9OhR2NjYkE4sEeCRTPyrYxrxaEJkyZIl+PLLLw027+rqajKHTWDBggUIDQ2FTCaDRqNBSEgIudDitWvXEBgYiNzcXERFRaF3797k7QYCNTU12LJlCz799FNSuzxEW3itU3h4OOzs7AwmQQwaNMhouzzGEgkiYg1B8R0EqqurkZKSgsrKSvEhn6pK4PHjx/j999+RlJQElUoFPz8/9OrVi8lYxNDQUJSWlsLT0xMSiQQSiYSkYkMfrVaLmzdv4sKFC6TijevXr4dEIkFRUREqKirg7u4ulnJSf4fnhfDwcLi7u8PHxweZmZlQKpX4/PPPSWw3NvJL0OuhguW9p8+lS5eYjkTkoaz+3Xffwd7ennwvB/77e9QYFHPk/xcoRs2xvqYEocujR4+K2VatVouzZ8+S6g8plUqkpqaivLwcDg4O6Ny5M5ycnMjsPy9je3ntIayFz69du4ajR49CpVJh9erViIiIwIcffkhmH2C7Vmq1GteuXYOLiwteeOEFxMbGAgD69u0LKysr0j0kPj4e0dHRyM7OhpWVFWxsbLBixQoS200FUyWCCRHWZWUCEokExcXFaNmyJVQqFWk7g8DRo0cRGBiIgwcPYsiQIdi+fTtppUBeXp747+LiYigUCjLbAjxEW1ivk0DdigeqHjseY4lYXJ8NsWzZMnTt2tUgaESFTCZD586d4e/vj5s3b2L37t0oKyvDu+++S34sHurErMQbWVWbPM88fPgQM2fOBFArzkUZcGadpRRgee8B7FuieCqrU2vB6POkAAIArFq1ivT6unfvnoH2QU5ODlxdXeHr62u0bdbXFOvydgD45ZdfoFar4enpiXv37iE3Nxepqalo37492UPf8zK2l/X5FmAtfH7gwAExsSeVSnHv3j1S+wDbtbKwsDCoQh02bJjB55R7yPHjx7F06VIsWbIECxcuxKZNm0jsNiVMQQQTIrwctqlTp2Lbtm2oqKiAtbU1pkyZQmofqM1YFBYWQiqVws/PD82aNSO1f+DAAfHfVlZWRpXD1SUlJQVAbUn1sWPHDKZYUPdbsV4nAUEgSaPR4Pr160hKSiJxQng4z126dIGjo6NB4IgF1tbW5HOMBRITE3HlyhWo1Wp06tQJ8+fPF+dlU2NjY4NTp06J5e2U+hQCrFTuhcxXWFiYQQBv3bp1ZNn15w0nJyccPHgQ3t7eyMjIIHUMJ0+e3GiWkhKW9x7AviWKh7K6sP81hQe6/5Vt27Y1OOL4nXfeMdo262uKdXk7AGRkZGDRokUAaqtNhH2xrjAeBU19bC/r8y3AelKNVCoVg1KsCtV5rRVrmjVrBqlUCplMhuLiYi6tnH81TEEEEyK8HDZPT896fa7UjB8/Hrt378Zrr70GtVpNklkA/jsbnXpN9BFEgOqKtlCLEQLs1kmfqqoqXL58GVevXkVaWhrGjh2LcePGkdjm4TzHx8dj3LhxBoEjAcqsj0QiwerVqw2CRlQil2vXroWHhweaN2+OuLg4xMfHi9cydaYyPDwcffr0Qffu3ZGRkYHw8HB89dVXpMfo1q2bgY7KnTt30KFDB6N7tZ8m/GSiPjNmzEBSUhIUCgXc3NxInUNeWUqW9x4AXLlypdHPKFoB9u7di4CAAGzevBmLFi0ycP6pKqka2v8EmtpsdB4jjllfUwIsy9vt7OwQGRkJT09P3Lp1C+7u7gBox03zQqVSwcHBAWq1GpcvX4a/vz/s7e3J+tp5nW/Wwufjxo1DWFgYsrOzsXTpUowdO5bUPsBvrVgzfvx4qNVqvPHGG9i6dStZC21TwhREMCHC2mETssaLFi1iXnIZFBQEFxcXlJSUQKVSkQlM7dixA5MmTUJYWFi9z3Q6HVxdXY1+kBF+1PR7HQEgOjraKLsNwWqdBJYvXw4LCwv07NkTU6dOxbfffks6o5mH8ywEPKZPnw6NRoPS0lLY29tDJpOR2BdgoaotUHcCR10oS0YrKytFUUJXV1cm1y0rlXveomjPA1KplJkQlwDrLCXLew9g3xLFQ1m9qQUKnoQwpjc6Opr090gf1teUAMvy9pkzZyIpKQlKpRJ+fn6ieOOMGTMA8NOooCA8PByLFi3Crl27YGdnhxMnTiA0NJRMA4r1+eYlfN6tWzd07doVZWVlzKoVed0brBGSWB06dCCbstPUMAURTNSDlcMmKPpOnTrVQJQpJyfnT9tsDFZjqIR+68bW4ciRI0Yfo6KiAmVlZbh48aLY26XVapGQkEDu8LAc1wUAvr6+SEtLQ2pqKpOpADycZ4HTp08jNjYWzs7OKCgowIABA5g5oLyhLBn18vLCpk2b4OPjg4yMDIN73VhYq9yXl5fD398fjx8/Fp0zrVaLzMzM/3Ojm0zQwbolioeyOk8aayeaO3cuiX3eI45ZwrK8vbEAoSCsSK1R0RBU4oo1NTUAgLKyMkyaNAnXrl0jscsL1pNqBK5fv45Tp04ZiB42hXaSurDeQ0zUYgoimODO9u3bG+xDpIT1GKqYmBjExsYaZJhCQkJISr9u3bqF5ORkFBQUiCWkMpmMSW8563WaMGECgFqRyKSkJGi1WmzYsAEvvvii2CdsDDyd55iYGCxZsgRSqRRarRYLFy4kDSL8/vvvAGrPiVKpJA/o8GLy5MnIyMhAXl4ehg4dSip4KWQQWancKxQKeHt7G8x9Bti0Epn468D63uPVEmVpaYmwsDCmyuos4TFHHgDc3NwAsB1xzGs/Z13ezgvWY3v9/f0RGhqKsWPHQq1Ww8LCgsSuAOvzzUv4/Ndff8W8efNIJ3DUheVa8dpDTNRiCiKY4AaPPkQBKysrREREMBlDBQAnT55EWFgYk9mzPXr0QI8ePVBYWAgLCwvY2tqSTTOoC+t1EmjVqhVGjx6N0aNHo6SkBMnJyQCAhIQE9OrVy2j7LJ1nISPdvn17pKenw8vLCwqFAu3btyexL1C3N3PVqlWk9nkRFxcn/vv+/fuiwCKF3gZrlXtBANTd3R3t27dn6kg1ddRqNfbv34/U1FRIpVLodDq0a9cOb7zxBrmD3hhUWUrW9x6vligeyuosRUd5tRP98ssvCA0NxdGjRzFv3jxS2wKsryle5e28WLBgAW7evIkDBw6Qju0VcHd3x8SJE8XX1EkZ1uebl/B5+/btYW5uTmqzLizXiscesm/fvkZ9ciFx9n8FUxDBxP+MsQ4bjz5EAZZjqAAgICAA9+/fNxgPRdX3WlZWhu3bt6OkpAT29vYoKSmBnZ0dJk+eTN6jxnqdGsLe3l48/ydPniQJIrB0nvX1LzIzM8ns1uXMmTPiv4uLi1FWVsbsWHWhnMd99epV2NjYiMEW4fo9deqU0Vkl1ir3AtXV1di3bx+KiopgY2MDX19fjBgxgukxmxpbt25Fly5dDJTsz58/j59++om8f551lpLXvce6JYqlsjoP0VFnZ2c4OzvDx8fHYI58bm4u6TSAF154ASEhIVAoFGJCAwCp0Czra4pXeTsvWI3tBWoTAWfOnEHfvn2h0+mg1Wpx+vRpg1GAxsL6fPMSPs/MzMSCBQtgZ2cHiUTCRHyZ5Vrx2EM6duwIoLbCzMXFBT4+PsjKykJubi6J/aaEKYhgoh6sHDaefYgsxx4BQFpaGtLS0gzeo/rh/vHHHzFq1CiDTPedO3ewZcsWzJkzh+QYAqzXiRcsnWdeDpnw90skEnh4eGDUqFHkx2D9MAbUim3pZyaXLVuGWbNmkSiGs1a5FwgKCoK9vT1SU1OhVCqRlpZmCiLUIT8/H7179zZ4r0+fPkyENFlnKXncewD7liiWyuo8RUdZzpEHID547d69m1lAkvU1xau8HQDWr18viikC/xXJpuwvZzW2NzY2FnFxcQa6SWZmZuQi0qzPN69JNV9//bXRNp4Gj/2W5R4itF4cPHhQ3Eu6dOnSoOD6844piGCiHqwcNm9vbwBs+xAFWI49Atg+WJaVldUrle/QocNTFfb/DKzXiRc8xhLJ5XIcOXIEKpWKieBQ//79cfHiReTm5sLc3JxJSTjrhzGg9sHi8OHD8PT0hEKhgJmZGbRaLUnZJWuVe4GPPvoIHTt2xNixY5lXPTRlysrKYGtrK74uLS1lchyWWUqA/b3HqyWKpbK6kOHr0KEDLl68iLy8PLI2pbrwmiM/ePBgbNmyBdXV1fjss88QExODwYMHk9hmfU3xKG/PyspCVlYWMjMzxeyxVquFXC4HQNtfzmpsb3BwMIKDg5Genk6qz1MXHr/fAPtJNdevXxf9HFa6KjzWisce0qJFC2zZsgVeXl6Qy+VMWrP/6piCCCbqwcphE3r31Go1AgMDkZubi6ioqHrZLApYjT1qaDylAFXJl4WFheigCeTm5jLRX2A5Hkqfe/fuGbR+5OTkwNXVlcwB5TGW6Mcff8SMGTOwefNmfPzxx4iJiSG1//3338Pd3R0+Pj7IzMxEeHg4Wa+xAOuHMQCYNWsWkpKSoFAo4OzsjFGjRkEqlWL+/PlG22atci+wfft2pKenIzU1FZGRkWjWrJlJ1bkO7733HlatWoWOHTvCyckJRUVF+P3338UpPJSwylIKsL73eLVE8VBW57FP8Zojv2nTJnz44Yf46aefIJVKcf78ebIgAut14lHeLpFIIJVKYWNjA5lMBp1OBwsLCybjPlmN7RWqJnbs2MF0tDiP+4IH+/fvZ66r8rzsIZ988gnS09ORn5+PQYMGMQ1S/VUxBRFM1IO1w3b06FEEBgbi4MGDGDJkCLZv347ly5eT2BZgNfao7hjBulCUlE2ZMgWbN2+Gg4MDHB0dUVRUBJVKhY8//tgouw3BcjyUPtu2bWtwIod+P7Ux8HCezc3N0bp1a2i1WrRu3bqegr+xPHz4EDNnzgRQO3+YRbUL63sbqA0S5ufno7y8HKNHj8a1a9fE+eLGwkvlPiYmBnfv3kV5eTmcnZ3JM8bPAz4+PggJCUFqaipUKhXatGmDCRMmMAl2sspSCrC+93i1RPFQVuexT/GaI6/VasVJDQBtKxzrdeJR3u7l5QUvLy8EBwdDqVSioqICOp0ODx8+NEhyGAPrsb2BgYEAan8f9Ks2qOFxX/CAZWuowPOyhxQWFuLMmTNiJdPp06fJgpBNBVMQwUQ9WDtsVVVVKCwshFQqhZ+fH5o1a0ZiV59nNfaIoqSMV+8bwH6deE3k4OE8BwYGQq1Wo2/fvpg7dy551NnJyQkHDx6Et7c3MjIyyHv5dDodPDw88Nprr9X7jHLMZ3h4OEaMGIFDhw5BJpPh2LFjZEEEXir3zs7O6Nu3L6ytrQ3ep7z3mjqZmZlwdXWFv78/VCoVjh07hitXruCVV14hvw9ZZSkFWN97Aqxbongoq/NaKx507NgRW7ZsQXFxMbZv3y5qPVDAa51Yl7cDwIoVK+Dk5CR+B8qRfKzH9u7duxcBAQHYvHkzFi1aZPBgTNke96zvCypxZB6toc96rajYuHEjs0qmpoIpiGCiHqwdtvHjx2P37t147bXXoFaryXsqdTodoqKi8Nlnn8Hd3b1JzcnWh7VzwGOdeE3k4OE8CwJiQ4cOFYX3KJkxY4bYBuDu7k7ezyeRSMReVpbU1NQgMDAQhw8fBsAmm8Fa5b5bt24Nvk/pmDd1tm/fLpYDf//99xg0aBDs7OywceNGktYVgH2WUoD1vSfAuiWKh7K6/lq5ubkxWSuWc+T1mTBhApRKJQICAuDm5mYw6thYeF1TPNDpdKSBZn1Yj+0dMmQIVq5caSCsKEC5l/M636zFkfVbQ1mNF+exVjz2EJaVTE0FUxDBhAgvhy0oKAguLi4oKSmBSqUiV8mVSCRwcHBAWloafHx8xE2QlyhbU4HHOvGayMHDeRbmoUdFReH69euwt7cncXIEpFIpWrduDRsbGwC1We9OnTqR2QdqhfBmz54NT09PcZ0ovwNQOz4tMjIS5eXliIqKgru7O6l9gL3KvYmnY2ZmBqlUirKyMqhUKlHbpqFWkz8L6yylAI97D2DfEsVSWT0lJUX8t42NDdq2bYv/x96ZR0V15O3/6W5AULZGWUSDLCqCE0VRXFABjcbE3cGMy+u4xF+icZugiRJUNoEor6PBncyM5lWjGBdMjElwQSWoIKgEBKMCDUFkka0F1Aa6f39w7h2aJS5UFXRzP+fknKb7pOpat6pu3W9963lEIhGVtqLpI98QTsejqqoKjx49QmJiIjGfd1Z9igWdO3fGwYMH1YIspNxwaNv2jhs3DuPGjcONGzcwfPhwKnUA7O43LXFkuVyOX3/9FQYGBhg8eDCOHj2KyspKTJ8+nXjWJYu2YjGH0Mxk0hSEIIIAD6sFG83UuKNHj2LOnDnIzMxEZmam2m8sdhBJpZTRhlU7cVFa2o4cLGyJOOX57Oxs+Pn5YcOGDUTLb25ckH6w0hDEasySJUuQnJyM0aNHw8rKiujZRFYq9wIvp1OnToiNjcX9+/f5xXldXR0UCgWxOmjvUnKwGHsA/SNRNJXVMzIy+M8ikQh1dXVISEhASUkJvvnmGyJ1cND0kW9IeHg4Jk2ahF69ehEvm1WfagmSa5GBAwcSK6sxrGx79fX1ERwcTM11gNX9piWOvGPHDowbNw7V1dX44osvsH79ehgYGCAiIuKlWmCvC4u2YjGH0Mxk0hSEIIIAD6sFG83UuPv37wOgHzCgnVL2Z5BYHLBqpyNHjiAwMBA//PAD1q1bR60eFrZEenp62LNnD+zs7KBSqYifw6+pqaE2LjgMDQ1x/vx5FBQUwNLSEuPHjydeB+cpzukgcOrYJGClct8SmhIkZMHq1atx5coV9O3blz/+VlZWRjQ1lfYuJQeLsQfQPxJFU1md29mrrq7GhQsXkJycDC8vLypngFn4yAP/1T6hAas+xWItQjP4wSpD9OTJk1RdB1jdb1riyCqVCu7u7gCAuLg43k2LhlAui7ZiMYfQzGTSFIQgggAPqwUbzdS4vLw87Nq1q9nfSAZCaKWUNYTm4oBVO3Xv3h3+/v7Iycnhj8gAIH7cgIUt0dq1a/HkyRNYWVmhtrYWy5YtI1q+l5cXtXHBERERAXd3dwwZMgSZmZmIiIjA+vXriZT9Mk9xErDSI8jJycHly5fV3D4++eQTJj7+q7UAACAASURBVEFCTcHAwAATJ07k/y4tLUVSUhKSk5OJZbKx2qVkMfYANkeiaCmrFxQU4Ny5cygsLMTYsWMREBBA5bw0wMZHHgAqKyuxbt069OzZk/+3kLofrPoUi7XI8ePHAdT3qby8PBgaGhLTPWFl20vbdYDV/aYlfF5WVsY/t+VyOf+5rKysVeU2B4u2YjGH0Mxk0hSEIIIAD6sFG83UOAsLCyp+0o2hlVLWEJqLA1btxHlWHzt2jGpwiuYCgcvQ2bdvH/8dpxdC8gUgJiYGgwYNorozU11dze+8WVtb48KFC8TKbugpzv0baHmK01a53717Nz788EOqbh/aQH5+PhISEvDw4UPk5uZiyZIlWLNmDbHyWe1Sshh7AP0jUTSV1VevXo0ePXrAzs4OiYmJuHnzJv8baV0VFj7yAIgHghvCqk+xWIs0nsO3b99OpFyAnW0vbdcBVveblvA5lyXV+DPniEQSFm3FYg6hmcmkKQhBBAEeVgs2T09P1NTUoKKignjZenp6TFKOaaWUNYTm4oBVO3GMGzcOkZGRvJ9ubGws0TRYmgsELy8vyGQyWFlZ4cGDB+jbty9cXV2b2P+1FkNDQ+oK3ra2tti3bx/s7e2RmZlJ9AxfQ09xWuObg7bKvY2NDRwcHKikcmoLn376KXr37o133nkHM2bMQGhoKPEAMatdShZjD6B/JIqmsnpLmWs0YOEjD9QHPk+dOkXlucSqT7FYizQU1SwtLSU6HlnZ9jYcG8bGxkTLBujfb9rC556enn/6e1hYGHx9fYnUxWJssJhDGmYyAaAiVN3eEVZIAjysFmzR0dFISUlBfn4+LC0toauri40bNxIp+//9v/8HoD46O3LkSCo7iSqVCjY2Npg5c2aT30ie86K5OGDRTg3Zt28fVT9dmovnu3fvIi4uDlOnTsXIkSNRWlqK48ePw93dnWggRiQSITw8HG+99RZ//aSzRRYtWoTMzEwUFhZi/PjxxIXdALrjm4O2yn1OTg4++eQTWFlZASB//EYbmDFjBm7duoVz587h8ePHqKurI14Hq11KFmMPoHckioWyOsugMysfeZo+76z6FK309oZwopoikQidO3fG2rVriZXNQdu2NyUlBefPn1c7okbyxZL2/WYlfN4SJAVzWYwNFnMIzUwmTUEIIgjwsFqwJSUlYfPmzQgICEBAQADR1Dhra2sAgKWlJQ4fPozq6moMGTIEw4cP5+1kWotIJCJ6zrs5aAcqWLRTQ2j56bJYPF+4cAG+vr78mTpra2v07dsXYWFhvBARCRoK/yiVSuIvxgBw5coV/vOjR494gcU+ffoQq4Pm+OagrXIfHh5OtDxtZMyYMRgzZgwUCgXu3LmDrl27YuvWrbC1tcUHH3xApA5Wu5S0xx7tI1EsldVZwMJHHqDr885iPgfopbc3ZNasWXj48CEKCgpgZWWFbt26ES0foG/b++2332LdunXUNkxYzSG0hc9ZwGJs0JxDbt++jUGDBiE1NbXJbzR0MNozQhBBgIfVgo1LEe7UqRPS09ORl5dHtHwAGDp0KIYOHYrKykpERkbi0KFDGDhwIN5//30iD9inT59izZo16NWrF0QiEfE0JhaBCoB+O3HQ8tNlsXgWi8VNRHn09PSIH//p27cvfvvtNyQnJ6OsrIxPkSPJ7du3YWhoyFsjVlRUwNjYGOfPnycWKGQxvmmr3JeUlOD06dMoLCyEpaUlpk+fTmXhrA3o6enBzc0Nbm5uqKurU0t9JgXtXUraY4/2kSiWyuosYOEjD9D1eafdp2intzfk4MGDePHiBezt7XHhwgVcvXqVmPsRK9teR0dH6OrqEi2zISzmEIC+8DkLWKx1WMwhrI6At2c08wkjQBXaC7aFCxeipqYG8+fPR0xMDObPn0+sbI7bt28jPj4eVVVVcHV1xccffwyVSoXw8HAiacksVNppByoA+u3EQctPl8XiuaCgAFFRUU3qLSgoIFL+tWvXkJycjBcvXsDFxQWPHz/Gpk2biJTdmGfPnqmJC4WGhsLHx4focQMW45u2yv2ePXswa9YsODg4IDMzE7t372bmDKHJSCQSKsK5tHYpWY092keiWCqrczuiLf1NAhY+8gCd5xKrPsUyvV0mkyEgIABAfeCW5FzIyrY3KysLGzZsgLGxMb+eIrHGYXW/WQmft3TElTsC2xpYrnVoziFnzpzBoEGD8Pvvv+Pjjz8mUqamIgQRBJpAM61MpVLh7NmzWLFiBXr27Eksmt2Y3NxczJ07t8lESGrAGxoa4vz583w6+Pjx44mU2xAWgQra7cRBy0+XxeJ5+fLlzX4/YMAAIuUfO3YMTk5OmDRpEuzt7dVUz0kjkUhw5swZ9OrVCzk5OdDR0YFSqYS+vn6ry87KyoJYLIatrS0AoGfPnhg7diy/00QS2ir3CoWCz8Tp168famtriZYv8GrQ3qVkNfZoH4liqazO7Yi29DcJWPjIy2QyGBgYwMbGBjKZDHfv3oWZmVmrj/Ox6lMs09v19fURFxcHOzs7ZGVlEXlecLAKzm7evJlKuazuN6td75aOuHJHYFsDy7UOzTlEX18fX331Fe7evdtEcFbTjpa0FpGKhmGqgEbCLdiOHDmCYcOG8Qu269ev4+9//zuxeg4fPoxhw4bB3t6eF1UhPUGWlpYiISEBVVVV/HckXlo5tmzZAnd3d17lPj4+HuvXrydWPlC/a9w4UNG5c2eiddBuJw5fX19MmjRJLVjh7Ozc6nIvX77c4m8vUxtuT+Tk5CAhIQHZ2dn4448/sHTpUirpl7W1tUhMTERxcTHMzc3h5uZGLGsjKCgIn3/+udoC8/nz59iyZQvxheKXX34JY2Nj2NjYYNKkSQgICCCaORMdHY309HTY2NggJycH/fv3V3s5E/gvL168QGpqqppgmYeHB5Gy/+yekupTLMZecHBws9k+LX1PGhLK6n92TIV0lsCvv/6Khw8fUvOR37VrF/T09PDs2TPI5XIMHToUnTt3RlxcHPz8/FpdPos+VV5eDlNTUxQXFzf5jbQIZnV1NS5cuMAf73rnnXeIr0Vo2/ampKTw5YeHh+PgwYPENrFY3O+ysrIWhc8tLS2J1cPBHXFNSUkhesSV1VqH9hxSWlqKI0eOYM6cOWpaKiwFaNsDQiaCAA/ttLKjR49izpw5yMzMRGZmptpvpF8ywsPDMWnSJPTq1YtouRzV1dW8P6y1tTUuXLhAvI6IiAi4u7tjyJAhyMzMREREBPFABe124qDlp8vSlogmvXr14u9BYWEhEhIScPLkSfj7++PZs2e8M0drUSgUKCoqQmVlJSZPnow7d+7A1dWVSNkqlarJDpW+vj5RsTIOWir3HNOnT4enpyeKi4sxefJkmJqaEi1fmwgJCcGgQYOoqF+z2KVkMfZoH4l6GSSU1TmFfqA+Nbiurg4JCQkoKSnBN9980+ryG0LbR76kpITvW+vWrcPEiRMBgJhVLIs+xSq9HQA6d+6McePGobKyEkD9UUvSQQTatr0nTpyAv78/QkJCIBaL8ccffxArm8X9ZiV8fuvWLVy7do3aEVdWax3ac4iZmRkWL16M8+fPo7KyEnPnzsXt27eFIIJAx4X2gu3+/ftM6gHovbRy2NraYt++fXwmAqkz/g1hEaig3U4cDf10uewTFmlfJG2JWGFpaYmpU6di6tSpAICtW7cSGzMRERGYOHEiTp8+DYlEgnPnzhELIujr6/Pq3Rz5+flEU19pq9zHxsbCy8sLx44da2ITSsOeTRvo0qULdc9v2ruUHLTGHu0jUSyYNWsWgP/uSicnJ8PLy4uoVS8HbR95LpiqVCqho6PDf6bxvKDVp1iKuu3fvx9PnjxRCxSSPm5J27ZXLBbzbUYzAZvW/WYlfJ6Xl4c5c+aga9euat/TOPtPc61Dew4BgJ07d/LrKbFYTHQ9pSkIQQSBJtBasOXl5TU5P8RB+oWy4UsrAOKihIsWLUJmZiYKCwsxfvx44hZzAJtABe124hD8dNsHNTU1cHFxwZkzZwCQXUwtWrQI+/fvh1QqhampKUpLS1FRUYGlS5cSq4O2yr2dnR0AzXq5ayu4QAsLz2/au5S0IXF0q60pKCjAuXPnUFhYiLFjxyIgIKBJoI0UtPtUjx49cOLEiWY/awoDBw5sMb2dNPn5+USPizUHbdveadOmITg4GHl5eQgJCeFfXDUN2sLnKSkpam3DiRiT0ERgCYvnEs31lKYgBBEEmkBrwWZhYcFsN6/xSyvpxc6VK1f4z48ePeJ1C/r06UOsDhaBCtrt1LDcU6dO4fnz51ixYgViY2Op7GAJ/Dndu3dHdHQ0KisrcfbsWaLWShYWFti4cSOePHnCn99saIuYkZEBJyenVtVBW+WeE4W8ePEiVq5cyX9/4MABrXgRJAkXaCFpi9cStHcptQWayuqrV69Gjx49YGdnh8TERDVRNNKBZ9o+8i/bReeOXrZnWKW3A4C7uzvOnTun9kJGWgeDtm3v4MGDMWjQIDx9+hRGRkbU1jq0oSV8npaWhrS0NLWjV3V1dVQcXlhAew4B6K6nNAUhiCDQBFoLNj09PWbnhbh6fv/9dyQkJODBgwdqmg+t5fbt2zA0NOTFJysqKmBsbIzz588Te4CzCFTQbieOvXv3YvHixfjXv/4FsViM+Ph4okEEmotnbWLJkiVITk7G6NGjYWVlhcmTJxOvo1u3bmrBA47jx4+3OqOJtsq9TCZDVlYWsrKyeKcPpVIJmUzW6rK1DS6osnPnTuoBF9q7lNoCTWX1lrIIOR4/fozu3bu3uh6AjY/8n8EdvWzPsEpvB4A7d+7A1NSUF2CmYblJy7ZXLpfj119/hYGBAQYPHoyjR4+isrIS06dP16h5hLZTjaWlJcRiMYqKivjAsEQi0VhBYRZzCIv1VHtHCCIINIHWgo17oWvphY8EdXV1SE1NRWJiIioqKvD48WOsXbuWqLsEUO+c8I9//IP/OzQ0FD4+PkSVtmkGKli1E4dSqVRLFSWd9kVz8cyKxkrt3KLqs88+I1YH98LHnds7cOAAFi1aRKx82ojFYj6AwKGnp0f0fLBYLIahoSFfpp6eHhO7VU2DZcCF9i4li7HHgqFDh2Lo0KG8svqhQ4eIKau/bAMgMjKy1UFClj7ytGHVp2intwP145q25SYt294dO3Zg3LhxqK6uxhdffIH169fDwMAAERERCAoKIlIHQP9+0xY+r6yshLOzM+rq6viAhVKpRFZWFvGAEc22YjGHpKen858NDAz4d6T09PQOl7EoBBEEmkBrwca90LX0wkeCJUuWYPjw4Zg1axbMzMwQGhpK5UVSIpHgzJkz6NWrF3JycqCjowOlUklURI5moIJVO3H0798fkZGRKCsrw4EDB4inQNNcPNPmZWmEJFSws7OzkZ2drfE77LRV7m1tbWFrawtPT0/U1NSgoqKCSLnaChdwkUgkUKlU1AIutHYpWYw9ltBWVqcNSx95WrDuU7TS2xvSuXNnHDx4kJpdHlAfrN2zZw/s7OygUqmIZVSoVCo+Sy0uLg5vvfUWABCzNmZ1v2kLkufk5MDOzq5J5jHJrBMWbcViDrl79y6A+jZ78eIF7OzsIJPJoKurKwQRBARoLdg4aL7wff7550hMTERkZCQcHBzw4sULQletjo+PDxITE5GTkwNzc3NMmjQJYrGYiMc0B81ABat24vD29kZubi7efvtt9OjRg7hIpCYvnptLI9TR0SGqLCwSidp8h53EUSZWKvfR0dFISUlBfn4+LC0toaurSzTLSBswMTGBp6cn+vXrR/18Ma1dShZjjyUsldVpEBERwfvIf/fdd3j06BHS0tKo+Mj/GePHj3/j/5dVn6Kd3t6QxvMrjfFOy7a3rKyMD5zL5XL+M6mz/qznEFrC52PGjIFSqcRf//rXVpfVEizaisUcwjnVfPnll2pr/rCwMCLlaxJCEEGgCbQWbBy3b99GfHw8lRc+JycnXrwtOzsbdXV1CAkJIR4I4SyiKisrMXnyZNy5c4e4tQvNQAWrdgLqH3oGBgawsbGBTCbD3bt3YWZmRiz7BNDsxXNaWhq8vLzw22+/ITU1lf/+9u3bxIRIWe6wBwUFYcCAARg5ciQsLCz470kELFhF+ZOSkrB582YEBAQgICAA27dvZ1KvJsEJu506darJb6SDU7R2KVmMPZa0pbJ642NGbwoLH3lunaFUKlFeXg5jY2NUVFRAKpUiMDAQI0eOfOOyWfUp2untDfH09AQA1NbWIiUlBYmJifDw8CBSNm3b3oZn+ht+5jQlWgvrOYSW8PmfaWGRyoJg1VYs5hAAePHiBX799Vc+E0ETLcVbixBEEGgCrQUbR25uLubOndtEE4H0C1+vXr1gamoKY2Nj4jZIERERvD+sRCKh4g/LIlAB0G2nXbt2QU9PD8+ePYNcLsfQoUPRpUsXfPXVV0SzNjTZlohzBGBhK8hih33Dhg1ITU3FyZMnUV5eDhcXF4wYMQKmpqZE66EJl+raqVMnpKenIy8vr42vqP3BLcInTJgAe3t7qr71tHYpWY49mrBSVpfJZEhKSkJ5eTmkUilcXV35NvT19SVaF0DPR54rIzIyEqtWrULXrl1RUlLSrNPB68KqT9FOb+d49uwZkpOTcfv2bTx48ABTp04l9gIO0Lft5QIgLREWFtaqvst6DqElfM6iP7XFfEtrDgGANWvW4NKlS7h37x4sLCywZs0aIuVqEkIQQaAJtBZsHKNHj0ZCQgKv9AvUp7uTfOGLi4vDzz//DAsLCxQVFeHdd98lWj4Lf1gWgQra7VRSUsJP2OvWrcPEiRMBgFj0XBtsib755hsEBAQgLi6OeuYEix12sVgMR0dHyOVyXLt2DRkZGXjw4AFMTEywYMEC4vXRYOHChaipqcH8+fMRExOD+fPnt/UltVtu3bqFU6dOQaVSwcbGBo6Ojhg8eDCRsmnvUrIcezRhoaweHx+PuLg4TJkyBVKpVM1elYQzSluQlZUFIyMjAIChoSGRHX3WfYpWejtQ/4Ktp6eHYcOGYcmSJdi+fTtxvQXatr0vo7W7x6zvNy3hc05kedOmTfxxFW6uJXUkVFvmWw5DQ0NMnToVV65cIZaZo2kIQQQBHtoLNo7w8HBMmjSJTzeiwS+//IKgoCBIJBLU1tbC398fY8aMIVY+C39YFoEK2u3EZVMolUro6Ojwn0mlfWmDLZG+vj6++uor3L17t4mNGumjJSx22Hfs2AGFQgE3NzesWrWKTx2MjIwkXhcNVCoVzp49ixUrVqBnz55YvHhxW19Su+aDDz5AXl4eMjIycPnyZWRkZBALItDepWQ59mjCQlmdtr1qWzBr1iwEBwdDLBZDpVLB29u71WWy7lO00tsBoE+fPnjw4AHu378PMzMzKmsQTe9XrO83LeFzzqWJpGNFY7Rlvm3M5cuXhSCCgADtBRuHubk5Ro0aRbRMDu68k0gkQllZGbp164by8nLiqbYs/GFpBipYtVOPHj1w4sSJZj+TgKUtES3Wr1+P0tJSHDlyBLNnz6ayUONgscM+ePBgtUDUvXv30K9fP+oWYaQQiUQwNTXFgwcPYG9vz+/K0EzX12QWL16M3r17Y/LkyQgMDCSmeg7Q36VkOfZowkJZnYW9KmtcXV0xcOBAyOVyYpbTrPsUrfR2AHxQpaioCImJiVAqldizZw+cnJzg5eVFpA5N71es7zdt4fPi4mJER0ejsLAQlpaWmDZtmpq2UWvQlvm2MYMGDWrrS2gzRCptuYsCrYZLV5w6dSpMTU1RWlqKc+fOEU9XDAoKQlVVFf9STDLTITAwEP7+/pDJZDh+/DiqqqrQpUsXfPDBB/x5LBLs3LkTK1eu5P/mUsFIk5ycjEePHsHa2hpDhgwhVi6rdnoZR48exZw5c974/798+TI8PT3x3XffqX0vEomI7CqxpKqqCufPn0dlZSXmzp2L27dvEzu+kpWVBbFYrHZvZTIZlEol7O3tidTBwfUtjv/93//F2rVridZBm+bSN1mdQdY0Hj9+jN9//x2ZmZmoqqqCsbExFi5cSKTswMBAtV1KoD67KSwsjOj9oDn2WMAFUJuD1MvY8uXLm2SpqVQqxMXFYffu3UTq4GjJR766upqoReLly5cRFxcHuVyOLVu2YMeOHfDx8SFSNqs+FR0djffffx9XrlxBTEwMevfuTTVVvKKiAjdv3sQ777yD69evY8SIEa0qj1W/Onv2LEaOHNkkWJSfn0/kCCer+x0UFIRNmzbxa9ANGzZg8+bNxMrfsGEDFi5cCFtbW8hkMhw8eJBo+QCbtqI5h6SlpbX4m6ZsXpFCyEQQ4GGVVtZYY4G0XRB3NpfGi0t2djays7ORlZXFWwUplUrIZDLidXEPCW5yJR2ooNlOr8r9+/db9f+zsCVixc6dO3kNDLFYTFQD4/Dhw/j888/VvrOyssKWLVuIvYzFxsYiNjYWubm58Pf3549COTg4ECmfJULA4NV5/Pgx8vPzUVxcDLFYDCsrK2Jls9qlpDn2WMBCWX3RokXNLr5JiqSx8JFvyKVLlxAUFITAwECIxWI8ffqUWNms+hSt9PaWMDEx4XURYmJiiAQRmoO0+J6lpSUOHz6M6upqDBkyBMOHD4ehoSExDShW95u28LmxsTEcHBwgEolgb2/Pa4aQhGZbsZhDMjIymv2eZOaXpiAEEQR4WC3YuDTU33//HQkJCXjw4MGfLoJeB5lM1uKZLhKLKZFIBLFYDENDQ75d9PT0iFqasQhU0G4nVrBYPLOCpgaGSqWCvr6+2nf6+vpE6/Dy8oKXlxeSkpKIZs20BY0zEXR0dGBhYYEpU6YQfUnWBoqLizFy5EjMnj2b+LOi4UKQQ6VSoaCggGg9LPRnaMJirvvxxx+p18PCR74hEokEz58/B1Cf4UKy/7LqU7TT22nDyrZ36NChGDp0KCorKxEZGYlDhw5h4MCBeP/999GvX79Wl8/qftMSPt+5cydEIhGePXuGNWvWwMbGBrm5uVSCUjTbisUcog2bVqQQgggCPLQXbHV1dUhNTUViYiIqKirw+PFjrF27Fn//+9+JlA/UW8jQXOjY2trC1tYWnp6eqKmpQUVFBfE6WAQqaLcTK7Th38BBUwNDX18fBQUFai/A+fn5TQILrYETZr1+/Tpu3Lih9psmLWoBoHfv3hg2bBif0nnt2jWMHDkSu3fvJhbw1BbeffddamWz2qVkIZRLExbK6lyZzS34Sb18s/KR55g3bx62bduGR48eYdu2ba06WtcYVn1KLpcDqN988PPzw4YNG6jUo+ncunUL165dQ1VVFVxdXfHxxx9DpVIhPDycyBihfb9pC5/Pnj271WW8KjTbisUcok2bV61F0EQQ4ElPT2/xNxLR4kWLFmH48OGYNWsWzMzMEBoaii+++KLV5TZkz549RF+2WyI6OhopKSnIz8+HpaUldHV11c5fkaJxoKJbt25EymXVTi+Dezl7U1gtnllBSwOjqKgI+/fvh1Qq5fVOKioqsHTpUmI2WuXl5TA1NUVxcXGT31hYdZGk4TlTlUqFjRs3qlljCmgftMaetrBo0aIW9XJILZyzs7NhZ2fX7FqExo51aWmp2hn5vLw8oi80LPrUl19+CWNjY9jY2GDSpEkICAgg/uz7448/8NZbb/F/czoC3377LebOnUu0Llp8//33cHd3R9euXdW+J6WJANC93+Xl5SgvL0dSUlIT4XOSz1eVSoW0tDSUlZXxAUMazgO02or1HNLREYIIAszIyMhAYmIiHj9+DAcHB6SlpVF70Wso2KdSqRAVFUU00sq9ZHAvFdu3b8enn35KrHyATaCCdjtx91epVKK8vBzGxsaoqKiAVCrVyJd8mrx48QKpqamorq6m9vB+8uQJysrKIJVK1QJSGRkZcHJyalXZ3NGb5hg7dmyrymZNTEwMrly5gm7duqG0tBRjxozBuHHj8NNPP2HSpEltfXkChGEx9lhAU1m9sWAqDbjnKWdXSJvQ0FAMGTIEY8eORXR0NHJzc4kJK7LqU7W1tWrp7dxnkjS+9//85z+JtRMrWhLaIwXt+81K+Pyf//wnrK2tkZCQgGHDhqG4uFhNRJwENNuKxRzScNOqMR1tXSscZxBghpOTE/+ikp2djbq6OoSEhFA5w9dQsE8kEuH3338nWj5nY9apUyekp6cjLy+PaPkAkJSU1CRQQRra7cQtPCIjI7Fq1Sp07doVJSUlOHnyJNF6aC6eWRESEoJBgwZBKpVSq6Nbt27NZrMcP3681S8ImmLJ9SpMmDAB77zzDuRyOYyNjfl/mxBA+C/aFCBkMfZY8NVXX6kpq0dERBBTVmeRTcTaR97X1xf79+/Ht99+C29vb6IvxrT7FO30dqBlsdzGu/ntmZcJ7ZGC9v1mJXwul8vh4+ODe/fuYfbs2diyZQuxsjlothWLOaQlPTEOEpsymoIQRBBoE3r16gVTU1MYGxujsLCQePl6enr47bff4OjoiPv37xP1LgeAhQsXoqamBvPnz0dMTAzmz59PtHyATaCCdjtxZGVl8Sq/hoaGyMrKIlo+zcUzK7p06UJNQIwFAwcOhFQqpTKeWZOSkoLvv/8e5eXlCA8Px3/+8x8sXry4rS+rXcEqQMgCTR97HDSV1YcPH87boz1+/Jh/IZs5cyZsbGyI1MHaR/7IkSOora2Fr68vTp48CX19fd55oLXQ7lNeXl6QyWSwsrJqkt5Osg4vLy9cuHCBWLuwpjmhPYlEwrtakIL2/WYlfC6RSFBTUwMzMzOcOHECpaWlRMsH6LYV6zmkOUhsymgKQhBBgDlxcXH4+eefYWFhgaKiIrz77rvEzqRxLFu2DNHR0fjxxx9hbW3dokDX65KVlQWxWMyfDe3ZsyfGjh37pz7dbwqLQAWtdmrMrFmzEBwcDLFYDJVKBW9vb6Lls7Aloo1IJEJ4eDjeeustPlWOhpgYLa5evYpp0h1BewAAIABJREFU06Y1+xLZHvQ3XocTJ07A398fISEhEIvF+OOPP9r6ktottAOELND0scdCWf3UqVN8YHbXrl1YuHAhjIyMsH//fqILZjMzMyxevLiJjzyNTAgnJyfeWs7X1xc//fQTsbJp96m7d+/y6e0jR45EaWkpjh8/Dnd3d2JtFRsbCy8vLzx58qSJ6LamjI/Kyko4Ozujrq6OX6cplUpkZWURteOjfb9ZOdV88cUXkEgk+Oijj5CSkkLlKCLttmI5h3R0hCCCAHN++eUXBAUFQSKRoLa2Fv7+/hgzZgzROp49e4YFCxZAJBLxE62pqWmryz18+DA+//xzte+srKywZcsWYgsploEKWu3UGFdXVwwcOBByuVxNyKq1sLQlos3kyZPbrG4SD9dp06YBqA8Y1NbWQi6Xw8TEhLiPNQvEYjG/wyPIBv05tAOELGjLsUcCFsrqXJZaaWkplEol+vTpQ60umj7yDbGxsUFkZCSeP3+OFStWoFOnTsTKpt2nWKS39+jRAwB5NxSW5OTkwM7ODvfu3VP7XiQSEQ0i0L7frJxqfvnlF7z//vvQ19eHq6srTpw4QXx+YTHfsppDOjpCEEGAGc+ePYOBgQFEIhHKysrQrVs3lJeXUzlL/fXXX2PTpk0A6h8WDf9uDSqVqoktnr6+PtEXDRaBCg5a7dSYy5cvIy4uDnK5HFu2bMGOHTuInD9laUtEG1tbW8TExKCyshJz5szBnTt3iNcRFBSEAQMGYOTIkWqaESQzBS5evIjLly/D3NwcxcXF8PDw0LhU2GnTpiE4OBh5eXkICQnB1KlT2/qS2i20AoQsYTH2aMIFAWkqq5uZmeH48ePIysriy1QoFKitrSVSfkNo+sg3ZN++fVi8eDH+9a9/QSwWIz4+HuPGjSNSNu0+xSK9/ciRIwgMDMQPP/yAdevWESuXJWPGjIFSqcRf//pXqvXQvt+snAV0dXURGhoKT09PXLx4kYqFL4v5ltUc0hwdKeNBCCIIMGPr1q3w9/fHhx9+iP/85z+oqqpCly5d8OGHHxKv68WLF/xnlUql9ndr0NfXR0FBgZr6cX5+fpPAQmtgEajgoNVOjbl06RKCgoIQGBgIsViMp0+fEimXxeKZFREREXzkXCKRUImcb9iwAampqTh58iTKy8vh4uKCESNGEM0+iY2NRVBQEMRiMZRKJTZu3KhxQQRzc3Ns2rQJT58+hZGRER4/ftzWl9RuoRUgZAmLsceC7du3N1FWJzUPrlixAnfu3IGzszO/g1tVVcUfs3v8+DG6d+9OpC6aPvINUSqV/G47QPZFg3afYpHe3r17d/j7+yMnJ4cXVgSgURbKwcHBLf5GclNGW+aQ0aNHIz09HVFRUfDw8ICbmxvxOli0FYs5hMWmTHtHCCIIMEWlUsHGxgZr166lWo+HhweCg4NhZ2eH7OxsYgupRYsWYf/+/ZBKpbzNTkVFBZYuXUqkfIBNoIKDVjs1RiKR4Pnz5wDqd69IZ5/QXDyzgkXkXCwWw9HREXK5HNeuXUNGRgYePHgAExMTLFiwoFVlc8dtHB0d8fDhQ9ja2iInJweOjo4kLp0p//nPf+Dv7w9jY2MAwLFjxzTuxZgVtAKELGnLXSuS0FRWF4vFGDx4sNp3UqmUV1iPjIwk9lK2ZMkSJCcnY/To0bCysqKW/ty/f39ERkairKwMBw4c4IX3SEC7T7FIb+fWNceOHdPYrD9WAnfaMods2bIFM2fOxOrVqxETE4OwsDD4+voSrYNFW7GYQ1hsyrR3hCCCADNkMlmL1iikJ/oJEyZg+PDhKCoqwtSpU/mXgdZiYWGBjRs34smTJygrK4NUKlWzzCNh7cIiUMFBq50aM2/ePGzbtg2PHj3Ctm3bMGfOHKLls7Alog2LyPmOHTugUCjg5uaGVatWwcDAAED9C0Brabjjo4nieoB2WJqxhnaAkAWsdr5pw0JZnQUvXryASqWCqakpqqqqcOXKFSpBYW9vb+Tm5uLtt9+GtbU1evXqRaxs2n2KVXo7AIwbN05NOyI2NpbYsQ/aHDhwAIsWLcKmTZt4ET9uXieZTaEtc8i6dev4DasJEyZQyaZg0VYs5hCamzKagkilqeEyAY0jMDCQSVQ4JSUF165dQ3l5OaRSKYYPHw4XFxfq9QJk/400AxUA23YqLS1VOy+dl5dH9MERHByM9evXY+/evbC2tsbNmzc1MpCQnJyMR48ewdraGkOGDCFe/tWrV9VETO/du4d+/foRr0fT0WRLM9Y8fPgQUVFRvIDZBx98AAcHh7a+rNeG9thjQV1dHR/USUlJQZ8+fZjpVJB89m3atKmJj7ynpyeRsgHgypUrLf5G8kVDG/oUUP985bQj/P39ERQUREU7SdPRhvtdUlKC8+fPqx0NpZGeT7utaM8hgPqmzLBhw9Q2ZT766COidbVXhEwEAWawEBs5c+YMioqKMHPmTN63/MyZM5DJZMR9gWnTrVs3teABBwkPWtbttG/fPgwZMgRjx45FdHQ0cnNziaaHs7AlYoGrqyvVc5SxsbFqQYSzZ88SDyLIZDJ8//33KC8v5xchmuaZ3KdPH/zf//0fqqqqqC6ktIFnz57Bz8+P/7uxCrqmQHvssYCFsnpLNBb5aw00feSB/6ZPp6SkwMDAAPb29sjOzkZ1dTXRIII29CmArnYEK4qLixEdHY3CwkJYWlpi2rRpaufYSaAN9zsiIgLe3t6IiorCBx98gNu3b1Oph3ZbsZhDbGxsMHPmzCa/dZQAAiAEEQQYwi3Cjx49yqezq1QqREVFEVvo3L59GwEBAfzflpaW+Oijj+Dv769xQQSasG4nX19f7N+/H99++y28vb2Jny9vy8Vza8nPz0dMTAw6d+6MESNG4Ntvv8Xz588xZcqUJmeQ35SW0vRp7Bh//fXXWLlyJfbv34+PP/4YsbGxxOugze7du/Hhhx9qrNsAS06dOqV2lpxGYIoWLMYeS1goq5eWliIhIQFVVVX8d97e3kTPTdP2ked2JK9fv47Vq1fz34eGhra6bG3rUwBd7QhWfPXVV1i4cCFsbW0hk8kQERGBzZs3t7pcbbnfe/fuxbJlyyASifD222/jxIkTGDBgAK9bQAKWbUV7DhGJRJDJZMTK01SEIIIAc+7fv89/FolE+P3334mVrVKp+BckDqVSySxyrinWLqzb6ciRI6itrYWvry9OnjwJfX19ouniLBbPtNi3bx/+53/+B9XV1di8eTPCw8PRqVMnBAcHE3uwenl5wcvLC0lJSdTTLHV1dWFlZQWlUgkrKyuN3Jm2sbGBg4MDdHSER2RLsAxM0YLF2GMJC2X18PBwTJo0iah+QGNY+MgD9RoSp0+fhp2dHWQyGSQSSavL1LY+BahrR/To0QM2NjZtfUmvjbGxMRwcHCASiWBvbw8jIyMi5WrL/S4qKgJQb7+oUCjwl7/8BYGBgUQFvVm2FYs55OnTp1izZg169eoFkUgEkUiEFStWUK+3PSGskASYo6enh99++w2Ojo64f/8+0YX62LFjER4ejvfeew9SqRSlpaU4d+4c8fT2trR2IRGoYNVOHE5OTnzqmq+vL3766Sei5bNYPNNCIpGgb9++AAArKyte3FJXV5dYHadPn8aMGTNw/fp13LhxQ+030g89FxcXKBQKjBo1Cp999hl69+5NtHwW5OTk4JNPPuEdUjTJ0owVLANTtGAx9ljCQlnd3Nwco0aNIlpmY1j4yAOAj48Pbt68CZlMBktLSyIvHtrWp4D67JO7d++iqqoKjx49QmJiIry9vdv6sl6JnTt3QiQS4dmzZ1izZg1sbGyQm5sLExMTIuVry/3Ozs7mg8FchsbTp09haGhIrA6WbcViDhGOOApBBIE2YNmyZYiOjsaPP/4Ia2vrFq2K3gQPDw84ODjgxo0bvCjh3LlziUfOWVi70AxUsGonDhsbGzV1506dOhEtn8XimRZFRUWIioqCSqXCkydP+M/czgAJvLy8AIDJEQ/uOMz48ePh5uZGbLHGkvDw8La+BI3Bzs4Ox44doy7ERQMWY48lNJXVuZexyspKrFu3Dj179uQz2UgHIln4yAP1QpS6urqQSqVQKBSIj49vtSaCtvUpgE32CS1oP/O05X7b2dlR1y5i2VYs5hBDQ0OcP38eBQUFsLS0xPjx44mWrwkIQQQB5jx79gwLFiyASCSCSqVCQUEB0Zfvnj17wtvbGzk5OaiurkZlZSXS09OJWiKxsHahHahg0U4c+/bt49WdxWIx4uPjiVpEsbAlokXDIFpDj2/u87Nnz3jV3zfl1q1bLf5GOvtkx44d+Mc//oGzZ88iJSUFJiYmGpPid+zYMbUjPg0heZ5Sm2AlxEUDFmOPJVVVVYiOjqYS0GGpMcPCRx4AQkJCmii4txZt61MAm+wTWnCZmyqVCmlpaWpjg4SIpjbeb1qwbCsWc0hERATc3d0xZMgQZGZmIiIiAuvXrydeT3tGCCIIMOfrr7/m7YFEIpHa36T48ssvYWZmxi8ORCIR0ZfjhtYuq1atUrN2IQWLQAXtduKgre5Mc/FMm5e199atW1u9QyAWi1v1/78OcrkcQH16pJ+fHzZs2MCs7tbScGEj8GrQFOKiDYuxxxKaAR3uZSw4OBgbN27kv+eChiRh4SMP0FFw17Y+BYBJ9glttm/fDmtrayQkJGDYsGEoLi4mEkTQlvv92WefUa+DZVuxmEOqq6v54Jq1tTUuXLhAvI72jhBEEGDOixcv+M8qlUrtb1KoVCqqNiuDBw9Ws8q7d+8e+vXrR7ROFoEK2u3EQVvdWZN3Q1kwcOBASKVSFBYWUq9LT08Pe/bsgZ2dHVQqFRGxMlbQCKBpOzSFuAReDRbK6mlpaUhLS0NBQQGioqIA1B8HKCsrI1YHx5IlS5CcnIzRo0fDysqKmkgabQV3bWHZsmVtfQmtRi6Xw8fHB/fu3cPs2bOxZcuWtr6kdkXnzp3b+hKIwmIOsbW1xb59+2Bvb4/MzEyNFBxtLUIQQYA5Hh4eCA4Ohp2dHbKzs4n6MnN07twZBw8eVBvUJNO2Y2Nj1YIINGzNWAQqaLcTR0N1Z2tra2JnK1ksnrWBq1evYtq0aTh58mST30hnbKxduxZPnjyBlZUVamtrtWIBKtA8KpUKLi4u0NPTw6xZs1BZWYkuXbq09WV1OFgoq1taWkIsFqOoqIgPAkskEmrWybR95AF2LhCajkgkwqlTp3hNo9jYWKLHEVkgkUhQU1MDMzMznDhxAqWlpW19SQKUoT2HLFq0CJmZmSgsLMT48eM1UkS6tQhBBAHmTJgwAcOHD0dRURGmTp3KK7SSZODAgcTLBNjamrEIVNBqJ44rV640+U4mk0EmkxEJHrFYPGsD06ZNA1AfMKitrYVcLoeJiQnRLAHOAWLv3r38rh43PjQt9VXg1RCJRIiJiYGLiwsAEFXyFnh1WCirV1ZWwtnZGXV1dVAqlQDqj6llZWXhL3/5C5E6WPnIp6WlESurI7B3716qmkYs+OKLLyCRSPDRRx8hJSWFmhOVQNvCag4B1Ne3jx494gUW+/TpQ7Se9owQRBBgSkpKCq5du4by8nJIpVIMHz6cX4CSRC6Xw9HRkbjXOwtbM5aBClrtxMHpE6SkpMDAwAD29vbIzs5GdXU1kSACi8UzbZoLtHB4eHgQPat48eJFXL58Gebm5vyZ0HfeeYdI2SwdIATaF2FhYbCzs9O4lHCWY48mLJTVc3JyYGdnh3v37ql9LxKJiAURWPnIZ2RkNPs9iX+LtvSphtDWNGLBL7/8gvfffx/6+vpwdXXFiRMniDyrtPF+04JFW7GaQwDg9u3bMDQ0hK2tLXJyclBRUQFjY2OcP39eYzS5WosQRBBgxpkzZ1BUVISZM2eia9euKCkpwZkzZyCTyYinRNrb2+Pu3bv44YcfUFtbix49emD+/PmtLpfbbb1+/Tpu3Lih9hup3VaW/uu02onD09MTAHD9+nWsXr2a/z40NJRI+SwWz7R5WaCF5FnF2NhYBAUFQSwWQ6lUYuPGjcSCCKamppDJZEhKSuKDhK6urrC1tSVSvkD7RJNTwlmOPU2Hm8t79uwJR0dHmJmZEa+DlY/8rFmz/vR37pjcm6CNfYq2phELdHV1ERoaCk9PT1y8eBHvvvsukXK18X7TgkVbsZpDgHo3iYaisqGhofDx8VETntV2hCCCADNu376NgIAA/m9LS0t89NFH8Pf3Jx5EMDIyQpcuXWBkZITa2lpiEwiL3VYWgQoOWu3UGIlEgtOnT8POzg4ymUyjxPZoQzvQAoBPP3Z0dMTDhw/5yLmjoyOxOuLj4xEXF4cpU6ZAKpWitLQUx48fh7u7O9zd3YnVI9C+0dHRQVlZGVHbPFqwGHssYLnb+fz5c3z33XcoLS2FoaEh+vTpg4kTJxIpm6WP/Muu403Rlj7VkIaaRj169NBIAbnRo0cjPT0dUVFR8PDwgJubG5FytfF+04JFW7GcQyQSCc6cOYNevXohJycHOjo6UCqVHeo4rRBEEGCGSqXiU/M5lEolldS4sLAw2NjYYOLEiXByciLmPXvr1q0WfyN1xo5lWjitdmqMj48Pbt68CZlMBktLS2K7l9qUKkgz0BIcHMx/zsrKIlZuQy5cuABfX1/o6ekBqLc86tu3L8LCwoQgghZz9uxZ1NXVoVevXsjNzeW/t7W11ZijLZoe5GS52+nm5gYTExPcv38fubm5ePDgAbEgAksfedpoep/ikMlkMDAwgI2NDWQyGe7evQszMzONOjIIAFu2bMHMmTOxevVqxMTEICwsDL6+vsTK15b7zQKabcVyDvHx8UFiYiJycnJgbm6OSZMmQSwWw8/Pj0j5moBIpYmHmwQ0kitXriAhIQHvvfcev1N57tw5DB8+nI9QkqS4uBhpaWn45ZdfoFQqsXXr1laXefny5RZ/I/VvuHTpUou/0RADotFOjXnx4gVSU1NRVVXFf0fDlUOTqa2txc2bN1FUVARzc3O4ublR0amgRWP/+Jd9L6AdfPnll1i/fj3/N7c49/PzQ0hISBte2auj6WOPJfPmzUP//v0xdepUYloIr0pgYCCT42sk6tGGPrVr1y7o6enh2bNnkMvlGDp0KDp37oy4uDiNe1F6/vy52g5xSUkJunbtSqx8bbjfrGjLtiI5h1RXVyMmJgaVlZWYM2cO7ty5Q91Rpr0h9HABZnh4eMDBwQE3btzg013nzp1LJTXu8OHDyMvLg46ODtzc3IilbQ8cOBBSqRSFhYVEymsOsVhMrezG0GqnxoSEhGDQoEEakeLcVigUChQWFqKyshKTJ0+m8kCSyWT4/vvvUV5ezmcAkXqgNvSP51CpVCgoKCBSvkD75MWLF7h27Rqf0qlQKABAoxbQLMaetnDgwAE8fPgQ9+/fR3R0NDp16qSxGWHl5eUwNTWFQqFAcnIynJ2dYWJi8lLNhFdBG/pUSUkJ/3xYt24dn3ESGxvblpf1RlRVVSE6OhplZWX8s4+k+J023G9WaEtbRUREYOLEiTh9+jQkEgnOnTunkf+O1sDubUVAAPWiTN7e3pgwYQL69++PyspKpKenEytfpVLh/PnzkMlkePHiBV68eAFDQ0M4OTkRKf/q1asAgJMnTzb5jxQDBw6Ep6cnnJycmvxHCtrt1JguXbpgxowZ8PT05P8TUCciIgK2trZ48OAB/0Aizddff40PPvgAKpUKH3/8MS9ARILly5fj7bffVvtvwIABaumFAtrHmjVrUFxcjJ9++glPnjzBmjVroFKp1LIT2jssxp62EBsbiwsXLiA9PR3m5uYYOnRoW1/SGxMREQEAOHr0KAoKCvDPf/4TAODs7EykbE3vUwqFAkVFRSgoKICOjg7/mQsUahIRERHo378/Hj16hFGjRqFLly7Ey9f0+80KbWmrmpoauLi48Bt/HTGxX3O2CgS0hi+//BJmZmb8rrRIJCLy0AaAQ4cOwdjYGOvWrYOuri5qamrw448/4tChQ1iwYEGry582bRqA+gh2bW0t5HI5TExMiJ7punr1KqZNm9ZsYIJU5Jx2OzVGJBIhPDwcb731lsbZwLGCeyCdOXMGAJ0Hkq6uLqysrKBUKmFlZdXErq01kBrDAppF586d0a1bN+jp6WHChAnIzs5G7969NebsOsBm7GkL5ubmzb6EZWRkUAtC06KmpgZAvS3wggULcOfOHaJla3qf6tGjB06cONHsZ02Bc9oQiUR4++23ceLECQwYMIC/L6TQhvvNCm1pq+7duyM6OhqVlZU4e/Ysevbs2daXxBwhiCDAHJVKhY8++ohK2dnZ2Wrp2bq6upg+fToCAwOJ1nPx4kVcvnwZ5ubmKC4uhoeHBzGrPBaBClbtxKHJNnCsYPFAcnFxgUKhwKhRo/DZZ5+hd+/exOsQ6FhERETA2dkZ8fHxeO+993D06FGN08AQFoOvTkte68ePH2/10SgWPvINcXJyQmBgIKZOnQqFQsGLwpJAG/rUyzYtjh49ijlz5jC6mjeDU+W3tbWFQqHAX/7yFwQGBhJX0NeG+80Kmm3Fcg5ZsmQJkpOTMXr0aFhZWXXIda4QRBBgTufOnXHw4EE1LQRSgoFKpbLJYkChUPAWd6SIjY1FUFAQxGIxlEolNm7cSCyIwEEzUMGqndLS0oiWp83MnDkT2dnZ/APJxcWFeB2cler48eN5lXUBgdbw9OlTTJgwAdevX2/rS3ljWIw9gZfDwke+IXPnzuU/P3z4kOgLRkfoU/fv32/rS3gp3IaJSqXC5s2bAdTPWaTdJTrC/SYFzbZiOYfs3LkTK1eu5HUQDhw4gEWLFhErXxMQgggCzBk4cCC1smfMmIGQkBB4eHjwDhBXr17FjBkziJTPvWQ7Ojri4cOHsLW1RU5ODhVBQpqBCtrtxJGRkdHs9yKRiLmyd3tn3759GDJkCN5//31ER0fj6tWr8PHxIVrHjh078I9//ANnz55FSkoKTExMsGLFCqJ1CHQsjI2Nce3aNSgUCiQmJsLU1LStL+m1YTH2BF4OCx/5hgQFBWHTpk04duwY5HI5SkpKiNn+CX2qfWBnZ8fE0UO4368OzbZiMYdkZ2cjOzsbWVlZvJuaUqmETCYjVoemIAQRBJgjl8vh6OgIBwcH4greLi4ucHBwwK1bt5CTkwOpVIo1a9bA2NiYSPnBwcH856ysLCJlNoZFoIJ2O3G8TOWaO68oAPj6+mL//v349ttv4e3tTWUBIpfLAdQ/BP38/LBhwwbidQh0LJYtW4ZLly7Bzs4OJSUlWLp0aVtf0mvDYuxpO+bm5sTKoukj3xBu17K4uBgrV64kegxH6FMdC+F+vzos2ormHCISiSAWi2FoaMiLKurp6RF1+9AUhCCCAHPs7e1x9+5d/PDDD6itrUWPHj0wf/58YuUbGRnBw8ODWHkNYRHRZhGoAOi206vCnVcUAI4cOYLa2lr4+vri5MmT0NfXJ35ERk9PD3v27IGdnR1UKhW1xblAxyEqKkpt/j558iT++te/tuEVvT4sxp62kJOTg8uXL6O6ulrNKo/kAtrHxwc3b95ETk4OLCwsqJ017tatG4KDg+Hh4YG6ujqi9sodoU+NHz++rS/hpbCyH+0I95sULNqK5hxia2sLW1tbeHp6oqamBhUVFcTK1jSEIIIAc4yMjNClSxcYGRmhtrYWurq6bX1Jr41MJsP333+P8vJyfiFFKsDAIlAh0P5wcnLiz9b5+vrip59+Il7H2rVr8eTJE1hZWaG2tlbIAhF4Y4qKilBYWIjU1FRe+6Surg6pqakaF0RgMfa0hd27d+PDDz+EmZkZtTpY+cgvX74cdXV1kEgkUKlUWLduHbGytaFPcULLSqUS5eXlMDY2RkVFBaRSKQIDAzFy5Mg2vsKXQ1pHoyW04X6zgkVbsZhDoqOjkZKSgvz8fFhaWkJXV1fjRIVbixBEEGBOWFgYbGxsMHHiRDg5OWmUFRjH119/jZUrV2L//v34+OOPERsbS7wOmoEKgfaHk5MTr1g8d+5cWFpaEiv79OnTmDFjBvbu3ctbbKpUKohEIkETQeCNePLkCe7du4eqqircu3ePz2xp72rtzUFz7GkbNjY2VI4iNiQiIgITJ07E6dOneR95GkGEtLQ0REVF8bpDs2bNwoABA4iUrQ19iltvREZGYtWqVejatStKSkqatZ/u6GjD/WYFi7ZiMYckJSVh8+bNCAgIQEBAALZv3060fE1ACCIIMGffvn0oLi5GWloaAgMDoVQqsXXr1ra+rNdCV1cXVlZWUCqVsLKywr1794jXwSJQwYry8nKYmppCoVAgOTkZzs7OMDExealmQkei4UNPLBYTfeh5eXkBAGbPnk2kPAEBZ2dnODs7Y8aMGVAoFKiurm7rS3pjaI49bSMnJweffPIJrKysANSfDyZtDczKR/7YsWPw8/ODgYEBqqurERISQiyIoE19KisrC0ZGRgAAQ0NDqscsNRVtut+0YdFWLOYQLpDaqVMnpKenIy8vj3gd7R0hiCDAnMOHDyMvLw86Ojpwc3Oj4mxAGxcXFygUCowaNQqfffYZevfuTbwOFoEKVkRERGDTpk04evQojI2N8fPPPyMwMBDOzs5tfWntBpoPPVNTU8hkMiQlJaG8vBxSqRSurq6wtbUlVodAx+Tf//43njx5AqlUyn+naQJTrF5atYHw8HDqddD0kW8Il40FgKgeAqBdfWrWrFkIDg6GWCyGSqWCt7d3W19Su0Ob7jdtWLQVizlk4cKFqKmpwfz58xETE0NU201TEIIIAsxQqVS4cOECZDIZ6urqUFdXB0NDQzg5ObX1pb0206dPB1AvLOTm5gYTExPidbAIVLCipqYGQL0/84IFC3Dnzp02vqL2B82HXnx8POLi4jBlyhTe0vP48eNwd3eHu7s7sXoEOh5FRUUafw6U1UurNlBSUoLTp0+jsLAQlpaWmD59Orp160a0Dpo+8g2ZPXvyG/u5AAAekUlEQVQ2QkNDIRKJoFKp8Le//Y1Y2drUp1xdXTFw4EDI5XKqWhiajDbdb9qwaCuac0hWVhbEYjG/CdOzZ0+MHTuWd1brSEgCAgIC2voiBDoGhw4dglgsxqJFizB27FiMGDEC6enpSExMpLZIoMWOHTswfPhwnD17Fj/88AN+++03uLm5Ea2jX79+kEgkcHBwwLBhwzT6Za+goACnTp3C+PHj0bVrVyQkJGDMmDFtfVntisGDB6OqqgpSqRTdu3fHpEmTiJV94MABfP755+jevTuMjIxgYWGBoUOH4v/+7/94X2UBgTchIyMDRUVFeP78OYqLi1FUVAQLC4u2vqzXgubY0za2bduG9957D1OmTIGRkREOHDhAfA7Zvn07pFIpJkyYgBs3buDatWsYMWIE0TpUKhVu3LiBZcuWwcvLC15eXkTPZmtTn7p8+TKOHDmCixcv4p133sH27duJ3w9NR5vuN21YtBXNOWTnzp1499131XRh9PX1sWfPng63niKbvyUg8CdkZ2dj+vTpvBuDrq4upk+fDplM1rYX9gbI5XIA9f8mPz8/FBQUEK9jx44dAICzZ89i165d2LVrF/E6WDF37lz4+/tj0KBByM3NZWa7pAlcuXKF/6+yshImJiaoqqrClStXiNUhFouhp6en9p2enh7xFF6BjkfXrl15ccWMjAyNOnbFYuxpGwqFAv369YOuri769euH2tpa4nX4+voiKysLS5Ysgb6+PhUfeZFIhD/++AMKhYJoudrYpy5duoSNGzfC0NAQYrEYT58+betLajdo4/2mBcu2ojmHqFQq6Ovrq32nr6/fIY+wCMcZBJihVCqhUCjUXmYUCoVGpgDp6elhz549sLOz41XJSdM4ULFhwwbidbAiKCgImzZtwrFjxyCXy1FSUgJfX9+2vqx2AffgSUlJgYGBAezt7ZGdnY3q6mp4eHgQqaOgoABRUVFN6qUR/BLoGHDztqbZOTaExdjTNlxdXREaGgobGxvk5ORQEY9j4SMP1ItELlu2DNbW1nxAtbUikdrYpyQSCZ4/fw6gfs0mBJ//izbeb1qwbCuac4i+vj4KCgp4cVkAyM/PbxJY6AgIQQQBZsyYMQMhISHw8PDgz2VfvXoVM2bMaOtLe23Wrl2LJ0+ewMrKCrW1tVi2bBnxOlgEKljBPTyKi4uxcuVKjT9DTRIu/e369etYvXo1/31oaCixOpYvX97s96SUyAU6HsHBwS3+pilWtCzGnrYxffp0eHp6ori4GJMnT4apqSnxOmj7yN+/fx96enoIDw/HoUOH+J11Ei8y2tin5s2bh23btuHRo0fYtm2bRtq40kIb7zctWLYVzTlk0aJF2L9/P6RSKUxNTVFaWoqKigosXbqUWB2aghBEEGCGi4sLHBwccOvWLeTk5EAqlWLNmjUwNjZu60t7ZU6fPo0ZM2Zg7969vKozp/C8YsUKonWxCFSwolu3bggODoaHhwfq6uqEnYxmkEgkOH36NOzs7CCTyYgGjQQXDAHSaEqg4FWgOfa0hdjYWHh5eeHYsWP8s4+DpCAhQN9HPioqij9S9+DBA6xatQq1tbX497//jf79+xOpQ5v6lJmZGfz8/Pi/O6KV3cvQpvtNGxZtRXMOsbCwwMaNG/HkyROUlZVBKpWqictmZGRopGD8myAEEQSYYmRkpNEpXl5eXgDqVZ1pwTJQwYrly5ejrq4OEokEKpUK69ata+tLanf4+Pjg5s2bkMlksLS0xOTJk9v6kgQEOgTC2Hs5dnZ2ANhkL9H2kVcqlXzq8YQJE/gXAJJHK7WpT+3btw9DhgzB2LFjER0djdzcXCo6FZqMNt1v2rBoK9pzCFC/OdacM83x48e1Ksj+ZwjbgQICr4GpqSlkMhmuXLmCM2fO4OrVq6iqqoK5uTmxOhoGKv72t7/hb3/7G/9ZU0lLS0NAQAD8/f2xadMmPHz4sK0vqd1RV1cHXV1dSKVSKBQKxMfHt/UlCQh0CISx93I4O7OLFy/C2dmZ/y8hIYF4XZyPPJexRlqwTCQS8Wf8R40aBQCorq4mGkTQpj7FQuhS09Gm+00bFm1Few4RqEfIRBAQeA3i4+MRFxeHKVOm8LoOx48fh7u7OzELRi5QkZSUhPLyckilUri6uvKLOE3k2LFj8PPzg4GBAaqrqxESEiKcx29ESEgIBg0aBKlU2taXIiDwynBZUi39rQkIY+/lyGQyZGVlISsrC5cuXQJQv3NPw12Jto+8t7c3tmzZgtGjR/PP8V9//RXe3t7E6tCmPsVK6FKT0ab7TRsWbUV7DhGoRwgiCAi8BhcuXICvry/vMGFtbY2+ffsiLCyMWBCBRaCCNQ1fLAQ9hObp0qWLRoqMCnRsvvnmGyxcuLDFvzUBYey9GmKxmLf5A+rFfz/55BPi9SxZsgTJyckYPXo0rKysiKc7Ozs7w8fHB7dv3+b1mT799FOi+kza1KdoC11qA9p0v2nDoq1ozyF/BsnM5PaOSCXkeAgIvDLBwcHNOgu09P2bEBgYqBaoAOptlcLCwjT2nFVqaiq+++47iEQiqFQqeHt7C5kIjdi6dStEIhHeeustPuCiyUdYBAQ0BWHsvR41NTWoqKjg/27uXPCb8Gd+8ZqmpaRNfaq4uBinT5/G8+fPsWLFCsTGxmLcuHFtfVntCm2637Sh2VYs55CgoCAMGDAAI0eOhIWFBdGyNQUhE0FA4DUoKChAVFSU2ncqlQoFBQXE6hCLxWoBBKB+x0dTd/BVKhWysrIQFBTU1pfSrhGEmAQ0iZ07d7Z4bEHTBGCFsffqREdHIyUlBfn5+bC0tISuri6xADpLH3naaFOf2rdvHxYvXox//etfEIvFiI+PF4IIjdCm+00bmm3Fcg7ZsGEDUlNTcfLkSZSXl8PFxQUjRoygYnvbXhGCCAICr8Hy5cub/Z7krjqLQAVLRCIR/vjjDygUiibBEYH/ItgwCmgSL3Ooefz4Mbp3787oalqHMPZenaSkJGzevBkBAQEICAjA9u3biZXN0keeNtrUp5RKJXr06MH/LSQwN0Wb7jdtaLYVyzlELBbD0dERcrkc165dQ0ZGBh48eAATExMsWLCAeH3tESGIICDwGrB4ULAIVLAmJycHy5Ytg7W1NUQiEUQiEQIDA9v6stoFzfmucwjpkALtlZed+4yMjGz3x6+Esff66OjULxs7deqE9PR05OXlEa+DhY88LbSxT/Xv3x+RkZEoKyvDgQMH8Pbbb7f1JbUbtPF+04JlW7GYQ3bs2AGFQgE3NzesWrUKBgYGAOqffR0FQRNBQEBAoA1JT09v8TdnZ2c8e/aMfzgJCGgKgYGB7T6IIIy910cmk6FHjx4oLCxETEwMBg8eDBcXF6J11NbW4ubNmygsLISlpSWGDh3KBy/aO9rap3Jzc/Ho0SNYW1ujV69ebX057QZtvd80YNlWLOaQq1evYsyYMfzf9+7dQ79+/YjW0d7RjFlZQEBAI7l//z709PRga2uLw4cPQy6XA6gXuOnfv38bX1374GXZLVu3bm33L2MCApqIMPZeD5VKhbNnz2LFihXo2bMnFi9eTKWe5nzkNUUTQZv6VHMidTKZDDKZTGPuB2206X7ThmVbsZhDYmNj1YIIZ8+eFYIIAgICAqSIiorCZ599BqA+oLBq1SrU1tbi3//+txBEEBDQYgT9E+1DJBLB1NQUDx48gL29PTXbXhY+8gIvR5uELgU6FjTnkNjYWMTGxiI3Nxf+/v68hbmDgwPxuto7QhBBQECAGkqlEvr6+gCACRMm8FZgSqWyLS9LQECAADKZDElJSSgvL4dUKoWrqytsbW0B1PvJC2gfmZmZyMzMVPuO9E4rCx95gZejTUKXAh0LmnOIl5cXvLy8kJSUhCFDhlCpQ1MQgggCAgLUEIlEeP78OfT19TFq1CgAQHV1tRBEEBDQcOLj4xEXF4cpU6ZAKpWitLQUx48fh7u7O9zd3dv68gQowSI1WyQSITw8nIqPvMDro8lClwIdE5pzyOnTpzFjxgxcv34dN27cUPtN0+yNW4sgrCggIECN9PR0fPfddxg9ejT/ovHrr7/C29tbOM7QiMuXL8PT0xMPHz7Ed999By8vLwwfPhzV1dXo3LlzW1+egIAagYGB8PX1VTu2oFAoEBYWpnFngIWx9+o0dtXR0dGBhYUFpkyZAisrKyJ1NCfApmkWetrUpzRZ6JIV2nS/acOirWjOIeXl5TA1NUVxcXGT317mWqRtkD3IJiAgINAAZ2dn+Pj4QEdHBzk5OdDR0cGnn34qBBCaIS4uDgDw888/Y+nSpYiOjgYAYQEi0C4Ri8VNdA/09PSIn49ngTD2Xp3evXtj3rx58PPzw7x58/DWW2/By8sLu3fvJlaHs7Nzk/80DW3qU82J1Amoo033mzYs2ormHHLr1i1cunQJqampTf7raAihRAEBAaoYGRmpKdgKNM/z58+Rnp4OfX19SKVSdOrUqa0vSUCgRQoKChAVFaX2nUqlQkFBQRtd0ZsjjL1XJyMjA/PmzQMAODg44ODBg/j73/9OJMWdpY88bbSpTwlCly9Hm+43bWi2FYs5RBMD5bQQgggCAgIC7YAFCxbg1q1b8Pb2Rk1NDYYNG9bWlyQg0CKLFi1qdudowIABbXA1rUMYe6/OmDFj4Ofnh27duqG0tBQeHh5QKpUYOnRoq8t+Wd8h6SNPG23qU4LQ5cvRpvtNG5ptxWIOGThwIKRSKQoLC1tVjjYgBBH+f3t3HhrV2bdx/DqTRWvNatLEBZtgrBqXutB0MXUBK0JcS6xSKBYExVSlSCNIqmmSRlGxiH2iqDSFUmq0Vu0GpahjaxcE61KC2qgxi2g0msbExJJl5vlDMu/ro7aT6ZxzZibfDwR0Bua+PMvtyW/uBQACwLVr17Rw4UJdunRJO3bs8KyMDQSib7/9NujWPngc7j3vTZ8+XdOmTVNTU5Oio6M938plZWX968+2ch95s4XSNcVCl/8slM632cw8Vlb0IT/++KPmzJmjL7744qH3cnJy/tVnBxvGZABAAPjfeYJffvmlzYmAv+d2u+VyuR76CTbce947e/asiouLVVRUJEkqLS21OVFgCqVraubMmcrKytKYMWM0evRojR492u5IASeUzrfZgv1YzZkzR9L9gsGSJUu0cOFCLV26tMcVECRGIgBAQGBOJYJJVVWVCgsLH/lesHxb3IV7z3v79+9Xfn6+iouL5XA4VFtba3ekgBQK11R5ebndEYJGKJxvq4TKsTpy5IiOHTumxMRE1dfXa/LkyZo2bZrdsSxFEQEAAsCiRYt0+vRpZWdnq62tjTmVCGgpKSlBVyx4HO497zkcDs8UBnYIf7xQuKbOnz//yNcNw9CoUaMsThPYQuF8WyVUjpXT6VRhYaEcDodcLpfWrl3b44oIYe+99957docAgJ4uLi5Ozc3NKi8vV1NTk6ZMmfLYVYYBu507d84vi+kFAu4978XExGj37t26evWqfv/9d82aNUv9+/f3axvHjh1TSkqKLl26pJ07dyoiIkKDBg1SRkaGIiIi/NqWWULhmho5cuQjf7rmne/YsSNk+oB/KxTOt1WsOFZm9iEul0tut1vXrl1Tnz59FBUVpStXrujevXt69tln/fQvCA6siQAAAeA///mP6urqlJqaqrq6Om3bts3uSMBjvfDCC2ptbZUkXb9+XVu3btXWrVtVU1Njc7Lu497zXmJiotatW6ctW7YoLy9PSUlJfm/Din3kzdYTrqmbN2/aHSFg9ITz7S9WHCsz+5CioiIVFRWpsrJSe/bs0YYNG/TZZ5/pypUr//qzgw3TGQAgANy+fVsrV66UJI0dOzZkhoojNB04cEDvv/++pPsPhW+++aaioqK0c+fOoLt2ufe8V1paqvz8fEVHR0u6vy/7qlWr/NpGKMyZ5prqWTjf3rPiWJnZh3Bu/w9FBAAIAPHx8Tpw4IBSU1N1+fJlxcXF2R0JeKzw8PuPDw0NDXK5XBo6dKjNiXzHvffPnE6nnE6nampqlJ+fL7fbLcMw1K9fP7+3ZeY+8lbhmupZON/es+JYWdGHVFVV6auvvlJjY6NnfZieVmCgiAAANnO73crMzFR7e7uqq6s1cOBAzZs3z+5YwGPFx8dr3759qqys1OTJkyVJbW1t6ujosDlZ93DveWfq1KmaOnWqDh8+bPriYWbuI2+FULumGhsbFRsbq7a2Nv32229KT09XTEyM5s+fb3e0gBBq59tMVh0rK/qQ3bt3a8WKFdq5c6eWLl0qp9Pp9zYCHUUEALCZYRg6cuSIVq9ebXcUwCvLly/XmTNnlJ6e7lmpvaWlRW+88Yak++sk+HvBPTNw73XP0KFD9cknn6ilpcXz7Zu/90c/fvy4pkyZ4pnPvHHjRr344ot+bcNMoXZNbdu2TevWrdOePXsUHR2t7777TgUFBZ4FFnu6UDvfZrLqWFnRh0RERCg5OVkul0vJycm6cOGCXz8/GFBEAIAAsWHDBqWmpnpWKl6wYIHNiYBHczgcGj9+/AOvxcXFeYam7tq1K6iGdnLveaekpESLFy9WfHy8aW2EwpoIUuhcU+3t7ZKk5uZmLVq0SGfOnLE5UWAKlfNtBbOPlRV9yNixY9XW1qbMzEzl5uYqLS3N720EOooIABAAZs6caXcEoEfi3vPe4MGDNWTIEM+aGGYIhX3kQ+maGjFihAoKCjR79my1tbUpMjLS7kgBJ5TOt9msOFZW9CFz586VJL3yyivKyMhQTEyM39sIdBQRAMBGFRUVioyMVHp6uj799FM1NTVJkmeeOQBzcO91X3V1tXJycpScnCzp/vDkgoICv7aRlpamW7duyel0KikpSTNmzPDr55spFK+p119/3fPnS5cuKTc318Y0gSUUz7dZrDxWVvQhW7du1dtvv61vvvlGZ8+eVUxMjJYvX+73dgKZw+4AANCT7d271/NAXlFRoddee02vvvqqZ19jIBgFw7eV3Hvdt3nzZu3atUuFhYUqLCz0ewFBsmYfebOE4jVVWFgo6f52nkePHtWWLVtsThQ4QvF8m8XKY2VFH9JVBLly5Yry8vJUV1fn9zYCHSMRAMBGLpdLvXv3liRNnz5dCQkJnteBQNfQ0KATJ06opaXF81p2drbWrFljYyrvcO95r6yszDN/+X/5ez6zFfvImyUUr6muBTTr6+u1YsUKrV271uZEgSMUz7dZrDxWVvQhkZGR2r59u1JTU+V2uxUWFub3NgIdRQQAsJFhGPrrr7/Uu3dvZWZmSpJaW1t5CEFQ2Lx5s7KysvT000/bHaXbuPe8N2bMGMvasmIfebOE4jWVkJCgoqIiTZ48WZ2dnXI4GMTcJRTPt1msPFZW9CHvvPOObt26peTkZHV0dGjZsmV+byPQGe6uEiMAwHLnzp3T559/rpdffllxcXFqaGjQTz/9pOzsbI0cOdLueMDf+uCDD7Rq1Sq7Y/iEey/wuN1unTp1Su3t7aqrq1NSUpIyMjKC5lu+UL2mOjs7FRYWJrfbrXv37qlPnz52RwoIoXq+zWDVsTK7Dzl48KDmzZunDz/80DM6y+12yzCMHrcmAkUEALBZc3OzTp8+rYaGBsXFxWncuHGKjo62OxbwWF0PUA0NDWppadGgQYM8D1TB9CDFvRd4Nm3aZPo+8mYKtWuqvLxce/fulcPhkMvl0vz58y0dmRLoQu18m8mqY2VmH9LY2KjY2FjV19c/9F5iYqIpbQYqiggAAKBbHvUA1aWnPUjBvzZt2qTOzk5T95GH9959913l5eXpiSeeUGtrq4qLi1VcXGx3LOCxzO5DqqqqdPLkSTU2NiouLk4TJkxQSkqK3z4/WLAmAgAA6JauQkFRUdEDC611bXsF+MqKfeThva6h2pJYDwFBwcw+5Oeff9bx48c1a9Ysz7SMffv2aeLEiZo4caJp7QYiiggAAKBbysvLVV5errq6Ou3du1fS/XnTf/75p83JEKys3Ece3lu4cKHWr18vwzDkdrsZFYKAZUUfcvjwYa1Zs8azjfGAAQP0zDPPaMOGDRQRAAAA/k5SUpIcDodu3ryp0aNHS5LCwsI0d+5cm5MhWO3du1e5ubmS7v8ysHLlSnV0dOijjz5ikTqbuN1uVVZWqrCw0O4owD+yog9xOByeAkKXyMjIHjlKhyICAADolrt37yo9PV2dnZ2eLbpcLpcqKys1atQom9MhGFm5jzy8YxiGamtr1dbW9tAvTkCgsaIP+f+j77q43W7V1dX5rY1gQREBAAB0S3V1tVJTU3XhwoUHXjcMgyICfGLlPvLwXnV1tZYtW6YBAwbIMAwZhqGCggK7YwEPsaIPeeuttx75ek/csYTdGQAAgE9+/fVXDRs2TPHx8XZHQZCzah95AKGJPsRaFBEAAIBPnE6nKioq1NDQoL59+2ro0KGaMWOG3bEQpKzaRx7/rGuRupSUlIcWqeMXMgQq+hDrMJ0BAAD4JCMjQzExMaqoqFBNTY0uXrxIEQE+i4qK0qRJk+yOAbHQJYITfYh1KCIAAACfLFmyRCNHjtTs2bO1cOFCu+MA8BMWugTwdygiAAAAn3z88ce6dOmSKioqdOjQIfXq1cvz7SWA4MVClwD+DkUEAADgE6fTqT/++EN3795VYmKihg0bZnckAH6QnZ2tjRs3PnKROgBgYUUAAOCTU6dOadiwYXryyScfeP38+fMaMWKETakA+AOL1AF4HIoIAADArwoKCpSfn293DAAAYAKH3QEAAAAAAEBwoIgAAAAAAAC8QhEBAAD4VWJiot0RAACASVgTAQAA+KS6ulrHjh1Ta2uruh4ncnJybE4FAADMxBaPAADAJyUlJVq8eLHi4+PtjgIAACxCEQEAAPhk8ODBGjJkiMLDeZwAAKCnYDoDAADwSW5uru7cuaPk5GRJkmEYKigosDkVAAAwE0UEAAAAAADgFcYfAgAAn9y+fVsHDx7UjRs3lJSUpLlz5yohIcHuWAAAwERs8QgAAHyyfft2ZWZmavXq1crMzFRJSYndkQAAgMkYiQAAAHzS1tam4cOHS5KGDx+ujo4OmxMBAACzUUQAAAA+mTBhgtavX6/BgwerurpaEyZMsDsSAAAwGQsrAgAAnzU2Nqq+vl6JiYmKjY21Ow4AADAZIxEAAEC3OJ1OTZ06VWVlZTIM44H3FixYYFMqAABgBYoIAACgW1JTUyVJY8aMsTkJAACwGrszAACAbklJSZEkHTlyROnp6Z6fEydO2BsMAACYjpEIAACgW6qqqlRZWanKykodPXpUkuRyuVRVVWVvMAAAYDqKCAAAoNscDof69u0rh+P+oMbIyEjl5OTYnAoAAJiN3RkAAIDP2tvbdefOHc/fExISbEwDAADMxkgEAADgk0OHDuns2bO6du2akpKSFBERobVr19odCwAAmIiFFQEAgE9Onjyp/Px89e/fX4WFherbt6/dkQAAgMkoIgAAAJ+Eh98f0NirVy+dO3dOV69etTkRAAAwG2siAAAAn1RVVWngwIG6ceOGvv/+e40fP15jx461OxYAADARRQQAANBtbrdbJSUlWr58ud1RAACAhZjOAAAAus0wDMXGxurixYvq7OyUy+WSy+WyOxYAADAZuzMAAACfXL58WZcvX37gtfz8fJvSAAAAKzCdAQAAAAAAeIWRCAAAwCcFBQUP/D08PFxPPfWUZs2apeTkZJtSAQAAM1FEAAAAPklLS9Pzzz+vlJQUVVVV6ZdfftFLL72kkpISFRUV2R0PAACYgIUVAQCAT86fP6+0tDSFh4dryJAhqqioUFpamsLCwuyOBgAATMJIBAAA4JNJkyYpLy9PCQkJamho0OTJk+VyufTcc8/ZHQ0AAJiEhRUBAIDPXC6XmpqaFB0dLYeDAY4AAIQ6/rcHAAA+OXv2rIqLiz3rH5SWltqcCAAAmI0iAgAA8Mn+/fu1Zs0azyiE2tpauyMBAACTUUQAAAA+cTgcnikMzI4EAKBnYE0EAADgk1OnTunrr7/W1atX9fTTTysrK0vjxo2zOxYAADARRQQAAOCT2tpaDRo0SM3NzYqKitL169c1YMAAu2MBAAATMZ0BAAD4pLS0VIZhKDo6WoZhqKyszO5IAADAZOF2BwAAAMHF6XTK6XSqpqZG+fn5crvdMgxD/fr1szsaAAAwGdMZAACATw4fPqxp06bZHQMAAFiIIgIAAPBJdXW1fvjhB7W0tHh2Z8jJybE5FQAAMBPTGQAAgE9KSkq0ePFixcfH2x0FAABYhCICAADwyeDBgzVkyBCFh/M4AQBAT8F0BgAA4JPc3FzduXNHycnJkiTDMFRQUGBzKgAAYCaKCAAAAAAAwCuMPwQAAN1SVlYmwzAe+d6CBQssTgMAAKxEEQEAAHTLmDFj7I4AAABswnQGAAAAAADgFYfdAQAAAAAAQHCgiAAAAAAAALxCEQEAAAAAAHiFIgIAAAAAAPAKRQQAAAAAAOCV/wKV1bDosLoWsQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 1296x432 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "diagramm = plt.figure(figsize=(18,6))\n",
    "plt.style.use('ggplot')\n",
    "\n",
    "if mitBerechnung:\n",
    "    x = x_train_ext_ohe.values.shape[1]\n",
    "    y = np.round(importances_ext[indices_ext], 5) \n",
    "else: \n",
    "    x = 37\n",
    "    y = [0.43254, 0.11091, 0.0912, 0.05419, 0.05144, 0.02158, 0.01528, 0.01416, 0.01289, 0.0125, 0.01117, 0.01037, 0.01028, 0.00965, 0.00955, 0.0093, 0.00771, 0.00751, 0.00736, 0.00736, 0.00678, 0.00668, 0.00613, 0.006, 0.00583, 0.00582, 0.00574, 0.00552, 0.00546, 0.00537, 0.0051, 0.0051, 0.005, 0.00492, 0.00473, 0.0045, 0.0044]\n",
    "    names_ext = [\"Duration_12_49\", \"AgeBand\", \"Disability_Category_Maternity\", \"OwnOccToAnyTransition_MOD_OwnOther\", \"OwnOccToAnyTransition_MOD_Own+1\", \"Disability_Category_Injury other than back\", \"Disability_Category_Other Musculoskeletal\", \"OwnOccToAnyTransition_MOD_Own+0\", \"Gross_Indexed_Benefit_Amount_1: $ < 1000\", \"Taxability_Benefits_Unknown/Invalid\", \"Disability_Category_Back\", \"Gross_Indexed_Benefit_Amount_2: $1,000 - $1,999\", \"Disability_Category_Mental and Nervous\", \"Disability_Category_Nervous System\", \"Disability_Category_Cancer\", \"Integration_with_STD_U: Unknown\", \"Gender_2:Female\", \"Gender_1: Male\", \"Integration_with_STD_I: Integrated with ASO of Fully-Insured STD\", \"Disability_Category_Digestive\", \"Gross_Indexed_Benefit_Amount_6: $5,000 - $9,999\", \"Disability_Category_Other\", \"Taxability_Benefits_T: 100% Taxable\", \"Integration_with_STD_N: Not Integrated with STD\", \"Gross_Indexed_Benefit_Amount_7:$10,000 - $14,999\", \"Gross_Indexed_Benefit_Amount_Unknown\", \"Gross_Indexed_Benefit_Amount_5: $4,000 - $4,999\", \"Disability_Category_Respiratory\", \"Gross_Indexed_Benefit_Amount_9:$20,000 And Over\", \"Disability_Category_Diabetes\", \"Taxability_Benefits_P: Partial Taxability\", \"Gross_Indexed_Benefit_Amount_4: $3,000 - $3,999\", \"Disability_Category_Circulatory\", \"Taxability_Benefits_N: Non-Taxable\", \"Gross_Indexed_Benefit_Amount_3: $2,000 - $2,999\", \"Gross_Indexed_Benefit_Amount_8:$15,000 - $19,999\", \"Disability_Category_Ill-defined and Misc Conditions\"]\n",
    "barlist = plt.bar(range(x), y)\n",
    "\n",
    "plt.xticks(range(x), names_ext, rotation=90, fontsize = 8)\n",
    "\n",
    "for bar in barlist:\n",
    "    yval = bar.get_height()\n",
    "    plt.text(bar.get_x() + 0.3, yval + 0.015, yval, rotation = 90)\n",
    "\n",
    "for i in range(x):\n",
    "    if names_ext[i].startswith('Taxability'):\n",
    "        barlist[i].set_color('g')\n",
    "    \n",
    "plt.title(\"Random Forest - Importance\")\n",
    "plt.ylabel('')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.04107,
     "end_time": "2020-08-30T09:34:14.237776",
     "exception": false,
     "start_time": "2020-08-30T09:34:14.196706",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Durch die Hinzunahme der weiteren Information konnte also keine wesentliche Verbesserung der Ergebnisse erzielt werden, so dass wir wieder zum ursprünglichen Modell übergehen, welches auch im Abschnitt mit den Vergleichsmodellen entwickelt wurde.\n",
    "Da wir nur die vorliegenden Daten zur Verfügung haben und diese bereits eine komprimierte und reduzierte Darstellung der Rohdaten darstellen (vgl. [Studie von 2008](https://www.soa.org/globalassets/assets/Files/Research/Exp-Study/research-2008-ltd-gltd-table.pdf)), können wir an dieser Stelle abschließend nur eine Verbesserung der Ergebnisse erzielen, indem wir ein Bootstrapping-Verfahren anwenden. Dabei handelt es sich um ein statistisches Resampling Verfahren, das in seiner einfachsten Form durch ein \"Ziehen mit Zurücklegen\" beschrieben werden kann.\n",
    "Die Idee ist, dass wir mit Hilfe dieses Verfahrens bessere Ergebnisse erhalten. Ein Blick in die Dokumentation des RandomForrestRegressor des packages 'sklearn.ensemble' verrät, dass bereits eine Option des Bootstrapping verbaut ist (vgl. [sklearn.ensemble.RandomForestRegressor](https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestRegressor.html)). Die standardisierte Ausprägung dieser Option ist mit dem Wert 'True' vorbelegt, so dass das zugehörige Bootstrapping-Verfahren bereits bei den obigen Varianten ausgeführt wurde. Wir können also an dieser Stelle nur untersuchen, welche Ergebnisse wir erhalten, ohne ein solches Verfahren auszuführen."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:14.329824Z",
     "iopub.status.busy": "2020-08-30T09:34:14.328732Z",
     "iopub.status.idle": "2020-08-30T09:34:14.336441Z",
     "shell.execute_reply": "2020-08-30T09:34:14.337420Z"
    },
    "papermill": {
     "duration": 0.058489,
     "end_time": "2020-08-30T09:34:14.337647",
     "exception": false,
     "start_time": "2020-08-30T09:34:14.279158",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "RF_boot_false: Erfolgreich ausgeführt\n",
      "RF_boot_false: Ergebnis mse = 0.012124425920309888\n"
     ]
    }
   ],
   "source": [
    "# RF in der Analyse ohne Bootstrapping\n",
    "if mitBerechnung:\n",
    "    regressor = RandomForestRegressor(n_estimators=250, min_samples_split = 150, max_features=3, bootstrap=False)\n",
    "    model = regressor.fit(x_train_ohe.values, y_train.values.ravel())\n",
    "    \n",
    "    predictions = regressor.predict(x_test_ohe.values)\n",
    "\n",
    "    mse_rf_false = ((predictions - datatest['Actual_Recovery_Rate'])**2).mean(axis=None)\n",
    "else:\n",
    "    mse_rf_false = 0.012124425920309888\n",
    "# Output\n",
    "output(\"RF_boot_false\", mse_rf_false, None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:14.427466Z",
     "iopub.status.busy": "2020-08-30T09:34:14.426400Z",
     "iopub.status.idle": "2020-08-30T09:34:14.432319Z",
     "shell.execute_reply": "2020-08-30T09:34:14.431209Z"
    },
    "papermill": {
     "duration": 0.052848,
     "end_time": "2020-08-30T09:34:14.432480",
     "exception": false,
     "start_time": "2020-08-30T09:34:14.379632",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.3501250902664061e-05\n"
     ]
    }
   ],
   "source": [
    "diff = mse_rf - mse_rf_false\n",
    "print(diff)"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.04166,
     "end_time": "2020-08-30T09:34:14.516917",
     "exception": false,
     "start_time": "2020-08-30T09:34:14.475257",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Wir sehen als Resultat, dass auch die Möglichkeit des Bootstrapping, auf den uns zur Verfügung stehenden Daten, keine signifikante Verbesserung des Vorhersagefehlers erreichen konnte.\n",
    "Abschließend lässt sich also feststellen, dass mit den gegeben Daten keine weiteren Verbesserungen der Resultate erzielt werden konnten. Oder anders formuliert, wir waren nicht in der Lage durch die Hinzunahme einer weiteren Information einen signifikant verbesserten 'mean squared error' zu erzielen. "
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.042012,
     "end_time": "2020-08-30T09:34:14.601327",
     "exception": false,
     "start_time": "2020-08-30T09:34:14.559315",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "<a id=\"AnalyseRF_Daten\"></a>\n",
    "## Hyperparameter Tuning\n",
    "Im vorherigen Abschnitt haben wir gesehen, dass wir aus den gegeben Daten keine Verbesserung des Vorhersagefehlers erzielen konnten. Da wir als Grundlage für unsere Analyse die gegebenen Daten verwenden wollen, können wir an dieser Stelle eine Verbesserung des Vorhersagefehlers nur noch durch eine bessere Kalibrierung des Modells erreichen.  \n",
    "Doch zunächst wollen wir kurz erklären, was sich hinter dem Begriff 'Hyperparameter Tuning', auf deutsch auch 'Hyperparameteroptimierung' genannt, verbirgt. Der Begriff Hyperparameter bezieht sich auf die möglichen Parameter des Trainingsalgorithmus und ist damit abzugrenzen von den Parametern, die während des eigentlichen Trainings des Modells anhand der Trainingsdaten festgelegt werden. Dieses Einstellen der Parameter können wir uns auch metaphorisch als Sendersuche bei einem Radio vorstellen. Das Verstellen der Frequenz, ist dabei bildlich gesprochen, das Einstellen eines Hyperparameter. Zunächst wählt man eine Ausgangsfrequenz und variiert nun schrittweise auf Grundlage des ertönenden Radioempfangs diese Frequenz. Dies wiederholt man bis man abschließend zu einen guten Ergebnis gekommen ist, oder alternativ, bis eine Veränderung der Frequenz keine Verbesserung des Radioempfangs nach sich zieht.\n",
    "Wie in der Dokumentation zum package 'sklearn.ensemble' beschrieben, besitzt die Methode RandomForestRegressor eine Vielzahl an möglichen Parametern (vgl. [sklearn.ensemble.RandomForestRegressor](https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestRegressor.html) zur Einstellung des jeweiligen Modells. Doch genau hier ist auch bereits das Problem. Wie soll man im Vorfeld nur auf Grundlage der Daten entscheiden, welche Einstellungen der Hyperparameter gewählt werden sollen, um direkt das bestmögliche Ergebnis zu erhalten? Dies ist schier unmöglich. Daher wechseln wir im Bereich des 'Hyperparameter Tuning' von der Wissenschaft zum Ausprobieren. Man probiert also eine Vielzahl von möglichen Kombinationen der Hyperparameter aus, indem man die Modelle auf Grundlage dieser Einstellungen kalibirert und den zugehörigen Vorhersagefehler misst. Doch bei diesem Ansatz haben wir ein Problem. Es ist recht einfach erklärt, denn wir trainieren unser Modell auf Grundlage der Trainigsdaten. Auch beim Hyperparameter Tuning benutzen wir nur die Trainigsdaten. Die Testdaten werden erst für die Vorhersagen benutzt. D.h. wir wählen die Hyperparameter gerade so, dass wir auf Grundlage der Trainingsdaten eine bestmögliche Vorhersage erzielen. Allerdings ist unklar, wie sich diese Wahl der Hyperparameter auf die Testdaten auswirkt. Es kann passieren, dass das Modell sehr gute Vorhersagen auf den Trainingsdaten durchführt, aber nur eine schlechte Performance auf den Testdaten zeigt. Dieses Problem nennt man  'overfitting'.\n",
    "In den bisher benutzen Modellen haben wir immer die Hyperparameter so gewählt, wie wir sie zum Vergleich mit den Modellen in der Programmiersprache 'R' benutzt haben. Wir werden an dieser Stelle dazu übergehen keinen der Hyperparameter vorzubelegen und finden in dem nachfolgenden Beispiel heraus, welche Hyperparameter im Standardvorgehen benutzt werden."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:14.693165Z",
     "iopub.status.busy": "2020-08-30T09:34:14.692281Z",
     "iopub.status.idle": "2020-08-30T09:34:14.785242Z",
     "shell.execute_reply": "2020-08-30T09:34:14.766091Z"
    },
    "papermill": {
     "duration": 0.1417,
     "end_time": "2020-08-30T09:34:14.785405",
     "exception": false,
     "start_time": "2020-08-30T09:34:14.643705",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Parameters currently in use:\n",
      "\n",
      "{'bootstrap': True,\n",
      " 'ccp_alpha': 0.0,\n",
      " 'criterion': 'mse',\n",
      " 'max_depth': None,\n",
      " 'max_features': 'auto',\n",
      " 'max_leaf_nodes': None,\n",
      " 'max_samples': None,\n",
      " 'min_impurity_decrease': 0.0,\n",
      " 'min_impurity_split': None,\n",
      " 'min_samples_leaf': 1,\n",
      " 'min_samples_split': 2,\n",
      " 'min_weight_fraction_leaf': 0.0,\n",
      " 'n_estimators': 100,\n",
      " 'n_jobs': None,\n",
      " 'oob_score': False,\n",
      " 'random_state': 42,\n",
      " 'verbose': 0,\n",
      " 'warm_start': False}\n"
     ]
    }
   ],
   "source": [
    "rf = RandomForestRegressor(random_state = 42)\n",
    "print('Parameters currently in use:\\n')\n",
    "pprint(rf.get_params())"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.050039,
     "end_time": "2020-08-30T09:34:14.882397",
     "exception": false,
     "start_time": "2020-08-30T09:34:14.832358",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Das sind vielleicht mehr Parameter als man zunächst erwartet hätte. Einige von den genannten sind bereits bekannt, andere unbekannt. Wie findet man nun die Hyperparameter, die einen entscheiden Einfluss auf die Vohersage der Modelle haben? Alternativ wäre es möglich mit allen diesen Parametern zu starten und somit auch herauszufinden, welche Parameter den entscheidensten Einfluss haben. \n",
    "\n",
    "* n_estimators = number of trees in the forest\n",
    "* max_features = max number of features considered for splitting a node\n",
    "* max_depth = max number of levels in each decision tree\n",
    "* min_samples_split = min number of data points placed in a node before the node is split\n",
    "* min_samples_leaf = min number of data points allowed in a leaf node\n",
    "\n",
    "Wir wählen für diese Parameter eine möglichst große Auswahl an Kalibirerungen die wir im Anschluss daran für eine Suche nach den besten Parametern benutzen wollen. Unsere Auswahl haben wir wie folgt getroffen:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:14.987414Z",
     "iopub.status.busy": "2020-08-30T09:34:14.986573Z",
     "iopub.status.idle": "2020-08-30T09:34:15.043329Z",
     "shell.execute_reply": "2020-08-30T09:34:15.030092Z"
    },
    "papermill": {
     "duration": 0.114197,
     "end_time": "2020-08-30T09:34:15.043463",
     "exception": false,
     "start_time": "2020-08-30T09:34:14.929266",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'max_depth': [10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110],\n",
      " 'max_features': ['sqrt'],\n",
      " 'min_samples_leaf': [1, 2, 4, 8, 16, 24],\n",
      " 'min_samples_split': [50, 100, 125, 150, 175, 200, 250, 300],\n",
      " 'n_estimators': [100,\n",
      "                  150,\n",
      "                  200,\n",
      "                  400,\n",
      "                  600,\n",
      "                  800,\n",
      "                  1000,\n",
      "                  1200,\n",
      "                  1400,\n",
      "                  1600,\n",
      "                  1800,\n",
      "                  2000]}\n"
     ]
    }
   ],
   "source": [
    "n_estimators = [100, 150, 200, 400, 600, 800, 1000, 1200, 1400, 1600, 1800, 2000]\n",
    "max_features = ['sqrt']\n",
    "max_depth = [int(x) for x in np.linspace(10, 110, num = 11)]\n",
    "min_samples_split = [50, 100, 125, 150, 175, 200, 250, 300]\n",
    "min_samples_leaf = [1, 2, 4, 8, 16, 24]\n",
    "\n",
    "random_grid = {'n_estimators': n_estimators,\n",
    "               'max_features': max_features,\n",
    "               'max_depth': max_depth,\n",
    "               'min_samples_split': min_samples_split,\n",
    "               'min_samples_leaf': min_samples_leaf\n",
    "              }\n",
    "pprint(random_grid)"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.049057,
     "end_time": "2020-08-30T09:34:15.142054",
     "exception": false,
     "start_time": "2020-08-30T09:34:15.092997",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Nun analysieren wir für jeden dieser Hyperparameter die Auswirkungen auf die zu erzielenden Ergebnisse. Dazu betrachten wir einen standardisierten Random Forest, bei dem wir einen der obigen Hyperparameter einbringen und entsprechend unserer Auswahl verändern."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:15.257541Z",
     "iopub.status.busy": "2020-08-30T09:34:15.256476Z",
     "iopub.status.idle": "2020-08-30T09:34:15.259876Z",
     "shell.execute_reply": "2020-08-30T09:34:15.259275Z"
    },
    "papermill": {
     "duration": 0.0683,
     "end_time": "2020-08-30T09:34:15.260013",
     "exception": false,
     "start_time": "2020-08-30T09:34:15.191713",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "def getPrediction_N_Estimators():\n",
    "    n_estimators = [100, 150, 200, 400, 600, 800, 1000, 1200, 1400, 1600, 1800, 2000]\n",
    "    if mitBerechnung:\n",
    "        train_results = []\n",
    "        test_results = []\n",
    "        for estimator in n_estimators:\n",
    "            rf = RandomForestRegressor(n_estimators=estimator, n_jobs=-1)\n",
    "            rf.fit(x_train_ohe.values, y_train.values.ravel())\n",
    "            train_pred = rf.predict(x_train_ohe.values)\n",
    "            mse_train = mean_squared_error(train_pred, y_train)\n",
    "            train_results.append(mse_train)\n",
    "            test_pred = rf.predict(x_test_ohe.values)\n",
    "            mse_test = mean_squared_error(test_pred, y_test)\n",
    "            test_results.append(mse_test)\n",
    "        print(\"done!\")\n",
    "    else:\n",
    "        train_results = [0.005342041264383930, 0.0053264808191175400, 0.005316773422128830, 0.0053020827200227100, 0.005298723872469110, 0.005294535676550300, 0.0052947801324029200, 0.0052940073100805800, 0.00529224527764862, 0.005291781739515370, 0.005291640969860730, 0.005290671961213600]\n",
    "        test_results = [0.015798919288405000, 0.015787426393295400, 0.015778847605019800, 0.01574739769698810, 0.015744049254123400, 0.015744072888633100, 0.01573609690736030, 0.01573560118532440, 0.015738821236662300, 0.01574254845739810, 0.01573572267187860, 0.015739351472915100]\n",
    "    return (n_estimators, train_results, test_results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:15.374345Z",
     "iopub.status.busy": "2020-08-30T09:34:15.369344Z",
     "iopub.status.idle": "2020-08-30T09:34:15.665800Z",
     "shell.execute_reply": "2020-08-30T09:34:15.666393Z"
    },
    "papermill": {
     "duration": 0.356932,
     "end_time": "2020-08-30T09:34:15.666556",
     "exception": false,
     "start_time": "2020-08-30T09:34:15.309624",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1296x432 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot - n_estimator\n",
    "plt.style.use('ggplot')\n",
    "\n",
    "(x, y1 ,y2) = getPrediction_N_Estimators()\n",
    "\n",
    "diagramm, ax1 = plt.subplots(figsize=(18, 6))\n",
    "\n",
    "color = '#E24A33'\n",
    "ax1.set_xlabel('n_estimators')\n",
    "ax1.set_ylabel('Train' , color=color)\n",
    "ax1.plot(x, y1, color=color)\n",
    "ax1.tick_params(axis='y', labelcolor=color)\n",
    "\n",
    "ax2 = ax1.twinx()  \n",
    "\n",
    "color = '#348ABD'\n",
    "ax2.set_ylabel('Test', color=color) \n",
    "ax2.plot(x, y2, color=color)\n",
    "ax2.tick_params(axis='y', labelcolor=color)\n",
    "\n",
    "plt.xticks(x)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.049606,
     "end_time": "2020-08-30T09:34:15.766213",
     "exception": false,
     "start_time": "2020-08-30T09:34:15.716607",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "In dem vorangegangenen Diagramm sehen wir den Einfluss des Parameters 'n_estimators' auf den durchschnittlichen quadratischen Fehler. Wir unterscheiden zwischen dem Fehler basierend auf den Trainings- und Testdaten. Auf den Trainingsdaten sehen wir, dass dieser mit steigender Anzahl an benutzten Bäumen sinkt. Dies gilt aber nur eingeschränkt für das Ergebnis auf der Testmenge. Basierend auf den Ergebnissen in der Studie nehmen wählen wir die Werte zwischen 150 und 1200 für unsere ausgedehnte Suche, dabei wählen wir als Beschränkung auf der rechten Seite den Wert 1200, weil sich im Anschluss keine wesentlichen Verbesserungen auf der Testmenge erzielen lassen.\n",
    "Im Wesentlichen steigt die Genauigkeit der Vorhersage mit der Anzahl der Bäume auf die der Random Forest zurückgreifen kann. Da aber auch die benötigte Zeit bei der Berechnung mit steigender Anzahl an Bäumen zunimmt, ist es sinnvoll hier diese Obergrenze zu wählen."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:15.880433Z",
     "iopub.status.busy": "2020-08-30T09:34:15.879426Z",
     "iopub.status.idle": "2020-08-30T09:34:15.882606Z",
     "shell.execute_reply": "2020-08-30T09:34:15.881862Z"
    },
    "papermill": {
     "duration": 0.066115,
     "end_time": "2020-08-30T09:34:15.882728",
     "exception": false,
     "start_time": "2020-08-30T09:34:15.816613",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "def getPrediction_Min_Samples_Splits():\n",
    "    min_samples_splits = [50, 100, 125, 150, 175, 200, 250, 300]\n",
    "    if mitBerechnung:\n",
    "        train_results = []\n",
    "        test_results = []\n",
    "        for samples_split in min_samples_splits:\n",
    "            rf = RandomForestRegressor(min_samples_split=samples_split, n_jobs=-1)\n",
    "            rf.fit(x_train_ohe.values, y_train.values.ravel())\n",
    "            train_pred = rf.predict(x_train_ohe.values)\n",
    "            mse_train = mean_squared_error(train_pred, y_train)\n",
    "            train_results.append(mse_train)\n",
    "            test_pred = rf.predict(x_test_ohe.values)\n",
    "            mse_test = mean_squared_error(test_pred, y_test)\n",
    "            test_results.append(mse_test)\n",
    "    else:\n",
    "        train_results = [0.010735437842069200, 0.011250717842914400, 0.011384832262341100, 0.01148265634233450, 0.01155735162963290, 0.011610059358956900, 0.01170601679840590, 0.011784996231501200]\n",
    "        test_results = [0.012157421798059200, 0.01206201528877350, 0.012041032456754100, 0.012037577469556200, 0.012035900145246700, 0.012029610615243300, 0.012036092993637500, 0.01205728898802250]    \n",
    "    return (min_samples_splits, train_results, test_results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:15.999630Z",
     "iopub.status.busy": "2020-08-30T09:34:15.998791Z",
     "iopub.status.idle": "2020-08-30T09:34:16.266980Z",
     "shell.execute_reply": "2020-08-30T09:34:16.266245Z"
    },
    "papermill": {
     "duration": 0.334191,
     "end_time": "2020-08-30T09:34:16.267128",
     "exception": false,
     "start_time": "2020-08-30T09:34:15.932937",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1296x432 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot - min_samples_split\n",
    "plt.style.use('ggplot')\n",
    "\n",
    "(x, y1 ,y2) = getPrediction_Min_Samples_Splits()\n",
    "\n",
    "diagramm, ax1 = plt.subplots(figsize=(18, 6))\n",
    "\n",
    "color = '#E24A33'\n",
    "ax1.set_xlabel('min_samples_split')\n",
    "ax1.set_ylabel('Train' , color=color)\n",
    "ax1.plot(x, y1, color=color)\n",
    "ax1.tick_params(axis='y', labelcolor=color)\n",
    "\n",
    "ax2 = ax1.twinx()  \n",
    "\n",
    "color = '#348ABD'\n",
    "ax2.set_ylabel('Test', color=color) \n",
    "ax2.plot(x, y2, color=color)\n",
    "ax2.tick_params(axis='y', labelcolor=color)\n",
    "\n",
    "plt.xticks(x)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.050368,
     "end_time": "2020-08-30T09:34:16.369358",
     "exception": false,
     "start_time": "2020-08-30T09:34:16.318990",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Hier lässt sich deutlich erkennen, dass der durchschnittliche quadratische Fehler auf der Trainingsmenge mit Zunahme der Werte deutlich ansteigt. Auf der Testmenge hingegen haben wir zunächst eine Verbesserung, die am Ende der Skala allerdings wieder in einem ansteigenden quadratischen Fehler resultiert. Daher wählen wir für unsere weiteren Betrachtungen die Menge beginnend bei 125 bis 250, unter Hinzunhame des weiteren Werts 225."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:16.484913Z",
     "iopub.status.busy": "2020-08-30T09:34:16.484014Z",
     "iopub.status.idle": "2020-08-30T09:34:16.486741Z",
     "shell.execute_reply": "2020-08-30T09:34:16.487356Z"
    },
    "papermill": {
     "duration": 0.067533,
     "end_time": "2020-08-30T09:34:16.487516",
     "exception": false,
     "start_time": "2020-08-30T09:34:16.419983",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "def getPrediction_Max_Depths():\n",
    "    max_depths = [int(x) for x in np.linspace(10, 110, num = 11)]\n",
    "    if mitBerechnung:\n",
    "        train_results = []\n",
    "        test_results = []\n",
    "        for depth in max_depths:\n",
    "            rf = RandomForestRegressor(max_depth=depth, n_jobs=-1)\n",
    "            rf.fit(x_train_ohe.values, y_train.values.ravel())\n",
    "            train_pred = rf.predict(x_train_ohe.values)\n",
    "            mse_train = mean_squared_error(train_pred, y_train)\n",
    "            train_results.append(mse_train)\n",
    "            test_pred = rf.predict(x_test_ohe.values)\n",
    "            mse_test = mean_squared_error(test_pred, y_test)\n",
    "            test_results.append(mse_test)\n",
    "    else:\n",
    "        train_results = [0.01173107880286860, 0.006923784449400560, 0.005378237789328430, 0.00534491552510019, 0.005343823606724510, 0.005341499937044560, 0.0053433433386073000, 0.005347726404402700, 0.005345536377698570, 0.0053420950625811700, 0.00534580650768231]\n",
    "        test_results = [0.01211029258016290, 0.013950975900494700, 0.01571448576584070, 0.015822057286910500, 0.015811746817061800, 0.015805402334199400, 0.015803887445595800, 0.01578507846102060, 0.015802389077742700, 0.01582470513830860, 0.01582724380447090]    \n",
    "    return (max_depths, train_results, test_results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:16.605060Z",
     "iopub.status.busy": "2020-08-30T09:34:16.599606Z",
     "iopub.status.idle": "2020-08-30T09:34:16.895430Z",
     "shell.execute_reply": "2020-08-30T09:34:16.894837Z"
    },
    "papermill": {
     "duration": 0.357361,
     "end_time": "2020-08-30T09:34:16.895552",
     "exception": false,
     "start_time": "2020-08-30T09:34:16.538191",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1296x432 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot - max_depths\n",
    "plt.style.use('ggplot')\n",
    "\n",
    "(x, y1 ,y2) = getPrediction_Max_Depths()\n",
    "\n",
    "diagramm, ax1 = plt.subplots(figsize=(18, 6))\n",
    "\n",
    "color = '#E24A33'\n",
    "ax1.set_xlabel('max_depth')\n",
    "ax1.set_ylabel('Train' , color=color)\n",
    "ax1.plot(x, y1, color=color)\n",
    "ax1.tick_params(axis='y', labelcolor=color)\n",
    "\n",
    "ax2 = ax1.twinx()  \n",
    "\n",
    "color = '#348ABD'\n",
    "ax2.set_ylabel('Test', color=color) \n",
    "ax2.plot(x, y2, color=color)\n",
    "ax2.tick_params(axis='y', labelcolor=color)\n",
    "\n",
    "plt.xticks(x)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.052049,
     "end_time": "2020-08-30T09:34:17.000152",
     "exception": false,
     "start_time": "2020-08-30T09:34:16.948103",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Der Hyperparameter 'max_depth' beschränkt im wesentlichen die Tiefe der Bäume innerhalb des Random Forest. Nach einhelliger Meinung sollte man diesen Parameter für eine möglichst aussagekräftige Bewertung nicht beschränken. Allerdings steigt mit der Tiefe der Bäume die benötigte Berechnungszeit, so dass es sinnvoll sein kann, in einigen Fällen eine Beschränkung vorzunehmen. Das vorangegangene Diagramm unterstreicht dies, da wir sehen, dass für größere Werte die Ergebnisse des Random Forest nicht wesentlich verändert werden.\n",
    "Wir werden dennoch diesen Hyperparameter in unseren Modellen nicht beschränken, da wir die Berechnungen auf einer anderen Maschine ausführen werden und die Dauer der Berechnungen in unserem Fall keine Rolle spielen."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:17.115917Z",
     "iopub.status.busy": "2020-08-30T09:34:17.115019Z",
     "iopub.status.idle": "2020-08-30T09:34:17.118254Z",
     "shell.execute_reply": "2020-08-30T09:34:17.117599Z"
    },
    "papermill": {
     "duration": 0.067101,
     "end_time": "2020-08-30T09:34:17.118384",
     "exception": false,
     "start_time": "2020-08-30T09:34:17.051283",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "def getPrediction_Min_Samples_Leafs():\n",
    "    min_samples_leafs = [1, 2, 4, 8, 16, 24, 32, 48]\n",
    "    if mitBerechnung:\n",
    "        train_results = []\n",
    "        test_results = []\n",
    "        for leafs in min_samples_leafs:\n",
    "            rf = RandomForestRegressor(min_samples_leaf=leafs, n_jobs=-1)\n",
    "            rf.fit(x_train_ohe.values, y_train.values.ravel())\n",
    "            train_pred = rf.predict(x_train_ohe.values)\n",
    "            mse_train = mean_squared_error(train_pred, y_train)\n",
    "            #mse_train = ((train_pred - y_train)**2).mean(axis=None)\n",
    "            train_results.append(mse_train)\n",
    "            test_pred = rf.predict(x_test_ohe.values)\n",
    "            mse_test = mean_squared_error(test_pred, y_test)\n",
    "            #mse_test = ((test_pred - y_test)**2).mean(axis=None)\n",
    "            test_results.append(mse_test)\n",
    "    else:\n",
    "        train_results = [0.005342041264383930, 0.00729622257790825, 0.009039979086927250, 0.010232387438883100, 0.010999459533349100, 0.01130805991928640, 0.011485263598531800, 0.011680837074353400]\n",
    "        test_results = [0.015798919288405000, 0.013723320992159900, 0.012733200110103300, 0.012317362048937700, 0.012136002458493700, 0.012073601092528600, 0.01204720230604530, 0.012044789719469200]    \n",
    "    return (min_samples_leafs, train_results, test_results)        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:17.233755Z",
     "iopub.status.busy": "2020-08-30T09:34:17.232952Z",
     "iopub.status.idle": "2020-08-30T09:34:17.503350Z",
     "shell.execute_reply": "2020-08-30T09:34:17.502759Z"
    },
    "papermill": {
     "duration": 0.33271,
     "end_time": "2020-08-30T09:34:17.503477",
     "exception": false,
     "start_time": "2020-08-30T09:34:17.170767",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1296x432 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot - min_samples_leafs\n",
    "plt.style.use('ggplot')\n",
    "\n",
    "(x, y1 ,y2) = getPrediction_Min_Samples_Leafs()\n",
    "\n",
    "diagramm, ax1 = plt.subplots(figsize=(18, 6))\n",
    "\n",
    "color = '#E24A33'\n",
    "ax1.set_xlabel('min_samples_leaf')\n",
    "ax1.set_ylabel('Train' , color=color)\n",
    "ax1.plot(x, y1, color=color)\n",
    "ax1.tick_params(axis='y', labelcolor=color)\n",
    "\n",
    "ax2 = ax1.twinx()  \n",
    "\n",
    "color = '#348ABD'\n",
    "ax2.set_ylabel('Test', color=color) \n",
    "ax2.plot(x, y2, color=color)\n",
    "ax2.tick_params(axis='y', labelcolor=color)\n",
    "\n",
    "plt.xticks(x)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.0514,
     "end_time": "2020-08-30T09:34:17.607699",
     "exception": false,
     "start_time": "2020-08-30T09:34:17.556299",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Das vorangegangene Diagramm zeigt den Verlauf des Hyperparameters 'min_samples_leaf'. Dieser gibt an, wie viele Datensätze in Blättern eines Knoten enthalten sein müssen, damit dieser als Knoten berückschtigt wird. Mit einem steigenden Verlauf dieses Hyperparameters wird also der entsprechende Baum nach und nach gestutzt und führt so zu einfacheren Bäumen. Daraus resultiert dann eine Glättung des Modells. In unserem Fall haben wir ziemlich viele Datenpunkte, so dass der Einfluss dieses Parameters nicht direkt ersichtlich wird. Wir sehen allerdings im vorangegangenen Diagramm, dass eine Steigerung des Parameters zu einem Anstieg des durchschnittlichen quadratischen Fehlers in den Trainingsdaten führt, während die Ergebnisse der Tastdaten verbessert werden. Die durchschnittlichen quadratischen Fehler der Trainings- und der Testdaten nähern sich an. Wir werden diesen Auswahl unverändert in unsere Suche nach den besten Parametern übernehmen.\n",
    "\n",
    "Zusammenfassend wollen wir die nachfolgenden Parameter benutzen, um die bestmögliche Wahl zu finden:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:17.719991Z",
     "iopub.status.busy": "2020-08-30T09:34:17.719008Z",
     "iopub.status.idle": "2020-08-30T09:34:17.743824Z",
     "shell.execute_reply": "2020-08-30T09:34:17.737515Z"
    },
    "papermill": {
     "duration": 0.084,
     "end_time": "2020-08-30T09:34:17.743967",
     "exception": false,
     "start_time": "2020-08-30T09:34:17.659967",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'max_features': ['sqrt'],\n",
      " 'min_samples_leaf': [1, 2, 4, 8, 16, 24],\n",
      " 'min_samples_split': [125, 150, 175, 200, 225, 250],\n",
      " 'n_estimators': [100, 150, 200, 400, 600, 800, 1000, 1200]}\n"
     ]
    }
   ],
   "source": [
    "n_estimators = [100, 150, 200, 400, 600, 800, 1000, 1200]\n",
    "max_features = ['sqrt']\n",
    "min_samples_split = [125, 150, 175, 200, 225, 250]\n",
    "min_samples_leaf = [1, 2, 4, 8, 16, 24]\n",
    "\n",
    "param_grid = {'n_estimators': n_estimators,\n",
    "               'max_features': max_features,\n",
    "               'min_samples_split': min_samples_split,\n",
    "               'min_samples_leaf': min_samples_leaf\n",
    "              }\n",
    "pprint(param_grid)"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.051879,
     "end_time": "2020-08-30T09:34:17.849805",
     "exception": false,
     "start_time": "2020-08-30T09:34:17.797926",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Diese Menge an Parametern wollen wir nun für eine Suche nach den besten Parametern berücksichtigen. Wir benutzen dazu das Tool 'GridSearchCV'. Dieses ermöglicht es uns das gewählte Grid zu übergeben und wird im Anschluss daran alle möglichen Kombinationen durchlaufen. Am Ende erhalten wir die Ergebnisse und können mit diesen die besten Parameter angeben.\n",
    "\n",
    "Anmerkung: Im Folgenden sind immer mal wieder einige Codezeilen auskommentiert. Dies liegt im wesentlichen daran, dass die Berechnungen mit dem 'random_grid' nicht in Kaggle durchführbar sind, da diese zuviel Zeit beanspruchen. Wir haben daher alle Berechnungen in lokalen Umgebungen laufen lassen und stellen hier nur die jeweiligen Ergebnisse vor."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:17.962401Z",
     "iopub.status.busy": "2020-08-30T09:34:17.961622Z",
     "iopub.status.idle": "2020-08-30T09:34:17.964914Z",
     "shell.execute_reply": "2020-08-30T09:34:17.964182Z"
    },
    "papermill": {
     "duration": 0.062694,
     "end_time": "2020-08-30T09:34:17.965059",
     "exception": false,
     "start_time": "2020-08-30T09:34:17.902365",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "rf = RandomForestRegressor()\n",
    "grid_search = GridSearchCV(estimator = rf, param_grid = param_grid, \n",
    "                          cv = 3, n_jobs = -1, verbose = 2, scoring=\"neg_mean_squared_error\")\n",
    "#grid_search.fit(x_train_ohe.values, y_train.values.ravel())\n",
    "#results = grid_search.cv_results_\n",
    "#results.to_csv('rf_grid.csv')\n",
    "#grid_search.best_params_\n",
    "#grid_search.to_csv('rf_grid_best.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.05225,
     "end_time": "2020-08-30T09:34:18.070301",
     "exception": false,
     "start_time": "2020-08-30T09:34:18.018051",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Mit dem obigen Grid dauert es sehr lange bis man die jeweiligen Berechnungen ausgeführt hat. Denn insgesamt haben wir an dieser Stelle 288 mögliche Kombinationen die geprüft werden sollen. Durch die 3-fache Kreuzvalidierung kommen wir auf insgesamt 864 nötige Fitting-Verfahren. \n",
    "Aus den Ergebnissen dieser Suche können wir nun die beste Kombination auswählen:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:18.182406Z",
     "iopub.status.busy": "2020-08-30T09:34:18.181425Z",
     "iopub.status.idle": "2020-08-30T09:34:18.205260Z",
     "shell.execute_reply": "2020-08-30T09:34:18.200480Z"
    },
    "papermill": {
     "duration": 0.082078,
     "end_time": "2020-08-30T09:34:18.205390",
     "exception": false,
     "start_time": "2020-08-30T09:34:18.123312",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'max_features': 'sqrt',\n",
      " 'min_samples_leaf': 1,\n",
      " 'min_samples_split': 125,\n",
      " 'n_estimators': 1000}\n"
     ]
    }
   ],
   "source": [
    "#best_random = rf_random.best_estimator_\n",
    "best_random = {'n_estimators': 1000, 'min_samples_split': 125, 'min_samples_leaf': 1, 'max_features': 'sqrt'}\n",
    "pprint(best_random)"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.052895,
     "end_time": "2020-08-30T09:34:18.311596",
     "exception": false,
     "start_time": "2020-08-30T09:34:18.258701",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Führen wir den Random Forest nochmals für diese Kombination aus, um den durchschnittlichen quadratischen Fehler für diese Kombination zu berechnen."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:18.429271Z",
     "iopub.status.busy": "2020-08-30T09:34:18.428366Z",
     "iopub.status.idle": "2020-08-30T09:34:18.435534Z",
     "shell.execute_reply": "2020-08-30T09:34:18.436129Z"
    },
    "papermill": {
     "duration": 0.071301,
     "end_time": "2020-08-30T09:34:18.436275",
     "exception": false,
     "start_time": "2020-08-30T09:34:18.364974",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "RF_best: Erfolgreich ausgeführt\n",
      "RF_best: Ergebnis mse = 0.012137927171212553\n"
     ]
    }
   ],
   "source": [
    "# RF_best\n",
    "if mitBerechnung:\n",
    "    # Zeitmessung starten\n",
    "    start_time = datetime.datetime.now()\n",
    "    \n",
    "    regressor = RandomForestRegressor(n_estimators=1000, min_samples_split=125, min_samples_leaf=1, max_features='sqrt')\n",
    "    regressor.fit(x_train_ohe.values, y_train.values.ravel())\n",
    "    \n",
    "    predictions = regressor.predict(x_test_ohe.values)\n",
    "    \n",
    "    mse_rf_best = ((predictions - datatest['Actual_Recovery_Rate'])**2).mean(axis=None)\n",
    "    \n",
    "    # Zeitmessung beenden\n",
    "    end_time = datetime.datetime.now()\n",
    "    \n",
    "    # verbrauchte Zeit berechnen\n",
    "    elapsed_time = end_time - start_time\n",
    "    t_rf_best = elapsed_time.total_seconds()\n",
    "else:\n",
    "    mse_rf_best = 0.01209225659854675\n",
    "    t_rf_best = 153.292926\n",
    "\n",
    "# Output\n",
    "output(\"RF_best\", mse_rf, None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:18.550137Z",
     "iopub.status.busy": "2020-08-30T09:34:18.548886Z",
     "iopub.status.idle": "2020-08-30T09:34:18.553832Z",
     "shell.execute_reply": "2020-08-30T09:34:18.554449Z"
    },
    "papermill": {
     "duration": 0.064214,
     "end_time": "2020-08-30T09:34:18.554601",
     "exception": false,
     "start_time": "2020-08-30T09:34:18.490387",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4.567057266580306e-05\n"
     ]
    }
   ],
   "source": [
    "diff = mse_rf - mse_rf_best\n",
    "print(diff)"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.05428,
     "end_time": "2020-08-30T09:34:18.662872",
     "exception": false,
     "start_time": "2020-08-30T09:34:18.608592",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Wir sehen also, dass wir durch das Hyperparameter Tuning eine Verbesserung des durchschnittlichen quadratischen Fehlers erreichen konnten. Wie auch in den vorangegangenen Abschnitten haben wir auch in diesem eine feste Trainings- und Testmenge benutzt. Es ist zu erwarten, dass bei einer zufälligen Auswahl dieser Mengen und einigen Wiederholungen des Prozesses mit Schwankungen bei diesem Ergebnis zu rechnen ist."
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.054546,
     "end_time": "2020-08-30T09:34:18.771270",
     "exception": false,
     "start_time": "2020-08-30T09:34:18.716724",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "<a id=\"ErweiterungNeuronalesNetz\"></a>\n",
    "# Erweiterung der Modellvielfalt um ein künstliches neuronales Netz\n",
    "Ziel dieser Arbeit ist unter anderem die Ergänzung der verwendeten Methoden um ein künstliches neuronales Netz, da diese in der Vergangenheit in unterschiedlichen Einsatzbereichen bereits sehr gute Ergebnisse erzielt haben. \n",
    "Das Vorbild künstlicher neuronaler Netze ist das menschliche Gehirn, das durch die Verknüpfung von Milliarden von Neuronen in der Lage ist, zu lernen und so auf neue Situationen zu reagieren. Ein künstliches neuronales Netz besteht deshalb ebenso wie das menschliche Gehirn aus Neuronen, die über gerichtete und gewichtete Verbindungen miteinander vernetzt sind. Während des Trainings des Netzes lernt es den Zusammenhang zwischen Input und Output, so dass es später auch auf unbekannte Daten angewandt werden kann. Das Netz lernt, indem die Verbindungsgewichte zwischen den Neuronen verändert werden.\n",
    "\n",
    "Es gibt eine Vielzahl unterschiedlicher Netze, die für verschiedene Anwendungsbereiche besonders geeignet sind. In diesem Fall haben wir uns für ein Feedforward-Netz entschieden, da diese einfache Form bereits ausreichen sollte, um den vorliegenden Zusammenhang abzubilden. \n",
    "\n",
    "Bevor das Netz für seinen konkreten Anwendungsfall verwendet werden kann, muss es mit Hilfe von Trainingsdaten trainiert werden.\n",
    "Dazu werden die kategorischen Daten als Dummy-Variablen benötigt und die Trainingsdaten müssen nochmal in Trainings- und Validierungsdaten gesplittet werden. Der Anteil wurde so groß gewählt, dass das Validierungssample in etwa gleich viele Einträge enthält wie das Testsample. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:18.888367Z",
     "iopub.status.busy": "2020-08-30T09:34:18.887088Z",
     "iopub.status.idle": "2020-08-30T09:34:18.911698Z",
     "shell.execute_reply": "2020-08-30T09:34:18.911096Z"
    },
    "papermill": {
     "duration": 0.086822,
     "end_time": "2020-08-30T09:34:18.911829",
     "exception": false,
     "start_time": "2020-08-30T09:34:18.825007",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "trainsample_nn= round(0.6*len(x_train_ohe))\n",
    "x_train_nn = x_train_ohe.iloc[:trainsample_nn, :]\n",
    "x_val_nn = x_train_ohe.iloc[trainsample_nn:]\n",
    "y_train_nn = y_train.iloc[:trainsample_nn, :]\n",
    "y_val_nn = y_train.iloc[trainsample_nn:]"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.053468,
     "end_time": "2020-08-30T09:34:19.019367",
     "exception": false,
     "start_time": "2020-08-30T09:34:18.965899",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Die Einstellungsmöglichkeiten für ein künstliches neuronales Netz sind vielschichtig. Die sogenannten Hyperparamter müssen a priori festgelegt werden. Sie umfassen beispielsweise die Anzahl der Hidden Schichten und die Anzahl der Neuronen auf den einzelnen Schichten. Wie schon beim Random Forest ist das sogenannte Hyperparameter Tuning von entscheidender Bedeutung.\n",
    "Um die Netzarchitektur zu optimieren, haben wir verschiedene Einstellungen von Hyperparametern mit Hilfe des Keras Tuners analysiert. Dabei werden die Einstellungen gewählt, die im Validierungssample das beste Ergebnis erzielt.\n",
    "Die folgenden Parameter haben wir getestet: \n",
    "* Anzahl Hidden Schichten: 1, 2, 3\n",
    "* Anzahl Neuronen: 17, 23, 64\n",
    "* Aktivierungsfunktion: ReLu, tanh, Sigmoid\n",
    "* Dropout-Rate: 0, 0.1, 0.2, 0.3, 0.4, 0.5\n",
    "* Lernrate: 1e-2, 1e-3, 1e-4\n",
    "\n",
    "Bei der Wahl der Neuronen-Anzahl haben wir uns an drei Faustregeln orientiert:\n",
    "1. Die Anzahl sollte dem Mittelwert zwischen Input- und Output-Größe entsprechen: (33+1)/2 = 17\n",
    "2. Die Anzahl sollte der Summe aus 2/3 der Anzahl der Input-Neuronen und der Anzahl der Output-Neuronen entsprechen: 22+1 = 23\n",
    "3. Die Anzahl sollte nicht mehr als doppelt so hoch wie die Anzahl der Input-Neuronen sein. Außerdem performen Potenzen von 2 tendenziell besser. Deshalb haben wir uns als dritte Möglichkeit für 64 Neuronen entschieden.\n",
    "\n",
    "Als Optimierer wurde der Adam-Optimierer gewählt. Die Einstellung der Dropout-Rate, der Anzahl der Neuronen und der Aktivierungsfunktion ist innerhalb eines Trials immer dieselbe für alle Hidden-Schichten. Das Training konnte sich maximal über 25 Epochen erstrecken. Das Training wird früher beendet, wenn drei Epochen hintereinander der Validierungsfehler nicht kleiner oder sogar größer geworden ist (Early Stopping mit Patience = 3). Die Batch-Size wurde aus Performance-Gründen auf 16 festgelegt. \n",
    "Aus diesen Einstellungsmöglichkeiten wurden dann insgesamt 100 zufällige Kombinationen getestet. Die Performance der einzelnen Netze wurde anhand des Mean-Squared-Errors des Validierungssamples gemessen.\n",
    "\n",
    "Die Funktion, die wir zum Ermitteln der bestmöglichen Netzarchitektur verwendet haben, ist untenstehend abgebildet. Der Aufruf der Funktion ist jedoch auskommentiert, da schon mit nur 100 Möglichkeiten und dem Verzicht auf Kreuzvalidierung, eine hohe Laufzeit resultiert. Wir haben den Code daher lokal ausgeführt und präsentieren wie schon beim Random Forest nur die Ergbenisse der Grid-Search."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:19.150261Z",
     "iopub.status.busy": "2020-08-30T09:34:19.149212Z",
     "iopub.status.idle": "2020-08-30T09:34:19.152508Z",
     "shell.execute_reply": "2020-08-30T09:34:19.151918Z"
    },
    "papermill": {
     "duration": 0.079069,
     "end_time": "2020-08-30T09:34:19.152632",
     "exception": false,
     "start_time": "2020-08-30T09:34:19.073563",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "def hypermodel(hp):\n",
    "    model = keras.Sequential()\n",
    "    # Tune die Anzahl der Units: Wähle zwischen 17,23 und 64 \n",
    "    hp_units = hp.Choice('units', values=[17, 23, 64])\n",
    "\n",
    "    # Tune die Aktivierungsfunktion: Wähle zwischen ReLu, tanh und Sigmoid\n",
    "    actfunc = hp.Choice('dense_activation',\n",
    "                    values=['relu', 'tanh', 'sigmoid'],\n",
    "                    default='relu')\n",
    "    model.add(keras.layers.Dense(units = hp_units, input_dim=33, activation=actfunc)) \n",
    "    \n",
    "    #Tune die Dropout-Rate: Wähle einen optimalen Wert zwischen 0 und 0.5\n",
    "    dropout = hp.Float('dropout', 0, 0.5, step=0.1, default=0.5)\n",
    "    model.add(keras.layers.Dropout(dropout))\n",
    "    \n",
    "    # Tune die Anzahl der Hiddenschichten: Ingesamt sind zwischen 1 und 3 möglich\n",
    "    for i in range(hp.Int('num_layers', 0, 2)):\n",
    "      model.add(keras.layers.Dense(units=hp_units, activation=actfunc))\n",
    "      model.add(keras.layers.Dropout(dropout))\n",
    "\n",
    "    model.add(keras.layers.Dense(1, activation='sigmoid'))\n",
    "\n",
    "    # Tune die Lernrate: zwischen 1e-2, 1e-3, 1e-4\n",
    "    hp_learning_rate = hp.Choice('learning_rate', values = [1e-2, 1e-3, 1e-4])\n",
    "\n",
    "    model.compile(optimizer = keras.optimizers.Adam(learning_rate = hp_learning_rate),\n",
    "                loss = 'mse',\n",
    "                metrics = ['accuracy'])\n",
    "\n",
    "    return model\n",
    "\n",
    "\n",
    "\n",
    "class ClearTrainingOutput(tf.keras.callbacks.Callback):\n",
    "  def on_train_end(*args, **kwargs):\n",
    "    IPython.display.clear_output(wait = True)\n",
    "\n",
    "\n",
    "def KerasTuner():\n",
    "    direc  ='RandomSearch_dir'\n",
    "    tuner = kt.RandomSearch(\n",
    "        hypermodel,\n",
    "        objective='val_loss',\n",
    "        seed=42,\n",
    "        max_trials = 100,\n",
    "        executions_per_trial= 1,\n",
    "        directory=direc,\n",
    "        project_name='RandomSearch'\n",
    "    )\n",
    "    epochs =25\n",
    "    # Funktionsaufruf für die Suche nach den besten Hyperparametern:\n",
    "    tuner.search(x_train_nn, y_train_nn, batch_size = 16, epochs = epochs, validation_data = (x_val_nn, y_val_nn),callbacks = [ClearTrainingOutput(), keras.callbacks.EarlyStopping('val_loss', patience=3)])\n",
    "    # Beste Hyperparameter:\n",
    "    best_hps = tuner.get_best_hyperparameters(num_trials = 1)[0]\n",
    "\n",
    "    # Printe Eigenschaften der Suche & der Ergebnisse: \n",
    "    tuner.search_space_summary()\n",
    "    tuner.results_summary()   \n",
    "\n",
    "#KerasTuner()"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.05381,
     "end_time": "2020-08-30T09:34:19.261815",
     "exception": false,
     "start_time": "2020-08-30T09:34:19.208005",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Das beste gefundene künstliche neuronale Netz hat die folgenden Einstellungen:\n",
    "\n",
    "* Anzahl Hidden Schichten: 2\n",
    "* Anzahl Neuronen: 64\n",
    "* Aktivierungsfunktion: Sigmoid\n",
    "* Dropout-Rate: 0\n",
    "* Lernrate: 1e-3\n",
    "\n",
    "\n",
    "Es ist im nachfolgenden Code abgebildet. Es erzielt einen MSE in Höhe von 0.0121."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:19.388630Z",
     "iopub.status.busy": "2020-08-30T09:34:19.383923Z",
     "iopub.status.idle": "2020-08-30T09:34:19.396403Z",
     "shell.execute_reply": "2020-08-30T09:34:19.395651Z"
    },
    "papermill": {
     "duration": 0.080656,
     "end_time": "2020-08-30T09:34:19.396529",
     "exception": false,
     "start_time": "2020-08-30T09:34:19.315873",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "FNN: Erfolgreich ausgeführt\n",
      "FNN: Ergebnis mse = 0.012136169058414551\n"
     ]
    }
   ],
   "source": [
    "def TrainFNN(num_units, act_func, learningRate, dropoutRate, num_layers): \n",
    "    if mitBerechnung: \n",
    "        model = keras.Sequential()\n",
    "        # 1. Hidden-Schicht\n",
    "        model.add(keras.layers.Dense(units=num_units, activation=act_func, input_shape=(33,)))\n",
    "        model.add(keras.layers.Dropout(dropoutRate))\n",
    "        # ggf. weitere hidden Schichten:\n",
    "        for i in range(num_layers):\n",
    "            model.add(keras.layers.Dense(units=num_units, activation=act_func))\n",
    "            model.add(keras.layers.Dropout(dropoutRate))\n",
    "    \n",
    "        model.add(keras.layers.Dense(units=1, activation='sigmoid'))\n",
    "    \n",
    "\n",
    "        opti = keras.optimizers.Adam(learningRate)\n",
    "        model.compile(optimizer=opti, loss= 'MSE')\n",
    "        callbacks = []\n",
    "        earlyStopping = keras.callbacks.EarlyStopping(monitor='val_loss', patience=3)\n",
    "        callbacks.append(earlyStopping)\n",
    "\n",
    "        model.fit(\n",
    "            x_train_nn, y_train_nn,\n",
    "            shuffle = True,\n",
    "            epochs=25,\n",
    "            validation_data=(x_val_nn, y_val_nn),\n",
    "            callbacks= callbacks,\n",
    "            batch_size= 16,\n",
    "            verbose = 0\n",
    "        )\n",
    "        ypred = model.predict(x_test_ohe)\n",
    "    \n",
    "        mse_fnn = mean_squared_error(ypred, y_test)\n",
    "    else:\n",
    "        mse_fnn = 0.012136169058414551\n",
    "    output(\"FNN\", mse_fnn, None)\n",
    "    return mse_fnn\n",
    "\n",
    "# Lege Parameter fest: \n",
    "num_units = 64\n",
    "act_func = 'sigmoid'\n",
    "learningRate = 0.001\n",
    "dropoutRate = 0\n",
    "num_layers = 1\n",
    "\n",
    "# Zeitmessung starten\n",
    "start_time =datetime.datetime.now()\n",
    "#Trainiere das Netz\n",
    "mse_fnn = TrainFNN(num_units, act_func, learningRate, dropoutRate, num_layers)\n",
    "# Zeitmessung beenden\n",
    "end_time = datetime.datetime.now()\n",
    "\n",
    "# verbrauchte Zeit berechnen\n",
    "elapsed_time = end_time - start_time\n",
    "t_fnn = elapsed_time.total_seconds()"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.054698,
     "end_time": "2020-08-30T09:34:19.505540",
     "exception": false,
     "start_time": "2020-08-30T09:34:19.450842",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "<a id=\"Fazit\"></a>\n",
    "# Zusammenfassung der Ergebnisse\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.054632,
     "end_time": "2020-08-30T09:34:19.614718",
     "exception": false,
     "start_time": "2020-08-30T09:34:19.560086",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Insgesamt verfolgte die Arbeit drei Ziele, deren Ergebnisse wir an dieser Stelle nochmal zusammenfassen möchten. \n",
    "\n",
    "Der Vergleich zwischen den Programmiersprachen 'R' und 'Python' zeigt, dass die Ergebnisse nur minimal voneinander abweichen. Beide Sprachen sind somit für die Arbeit im Rahmen der aktuariellen Welt geeignet. Gerade bei den GLM´s ist zu erkennen, dass die Ergebnisse der beiden Programmiersprachen identisch sind. Dies liegt im Wesentlichen an der identischen Programmierung innerhalb der Programmiersprachen, da beide im wesentlichen FORTRAN Bibliotheken benutzen. \n",
    "Die Wahl der Programmiersprache ist also nur abhängig vom persönlichen Geschmack der Entwickler.\n",
    "\n",
    "Der Random Forest ist eines der meistgenutzten Tools im Rahmen von Data Science. Wir konnten mit einigem Aufwand erreichen, dass mit einer optimierten Wahl von Parametern der durchschnittliche quadratische Fehler minimiert wurde. Es soll aber nicht verschwiegen werden, dass diese Arbeit nicht ohne Mühen verlaufen ist und es daher zu entscheiden ist, ob der Einsatz der Zeit sinnvoll erscheint. \n",
    "Wir sehen aber, dass durch das Hyperparameter Tuning die Ergebnisse verbessert werden konnten. Im Gegensatz zu den einfacheren Methoden, wie zum Beispiel eines GLM, kommt es beim Random Forest auf die Wahl der Parameter an. Mit der richtigen Kalibrierung kann am Ende das Maximum herausgeholt werden. Das geht dabei aber deutlich zu Lasten der Entwicklungszeit die benötigt wird, um diese optimale Einstellung zu finden.\n",
    "\n",
    "Das künstliche neuronale Netz ist ebenfalls ein vielfach eingesetztes Tool im Bereich der Data Science. Wir konnten zeigen, dass dieses Machine-Learning-Verfahren ebenso gute oder sogar bessere Ergebnisse erzielt wie die anderen bereits untersuchten Methoden. Wie beim Random Forest ist die Einstellung der Hyperparameter von entscheidener Bedeutung. Der Prozess des Hyperparameter-Tunings ist dabei jedoch nicht trivial und zeitintensiv. \n",
    "\n",
    "Insgesamt lässt sich feststellen, dass je nach Einsatzzweck überprüft werden muss, welches Modell angemessen ist. Das GLM weist den schlechtesten MSE der getesteten Modell auf - es lässt sich jedoch sehr schnell umsetzen, verstehen und gut interpretieren. Insbesondere beim Random Forest und dem künstlichen neuronalen Netz nimmt die Suche nach den richtigen Hyperparamtern viel Zeit in Anspruch, so dass für jeden Einsatzzweck individuell abgewogen werden muss, ob der Mehraufwand im Verhältnis zu der besseren Prognose steht. \n",
    "\n",
    "Abschließend geben wir noch einmal eine grafische Darstellung der einzelnen durchschnittlichen quadratischen Fehler aus dieser Arbeit an."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:19.737085Z",
     "iopub.status.busy": "2020-08-30T09:34:19.735980Z",
     "iopub.status.idle": "2020-08-30T09:34:19.739749Z",
     "shell.execute_reply": "2020-08-30T09:34:19.739043Z"
    },
    "papermill": {
     "duration": 0.069975,
     "end_time": "2020-08-30T09:34:19.739867",
     "exception": false,
     "start_time": "2020-08-30T09:34:19.669892",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "MSE_data = [('GLM1', mse_glm1 , t_glm1 , 'sec' ) , \n",
    "         ('GLM2', mse_glm2 , t_glm2 , 'sec' ) ,\n",
    "         ('GLM3', mse_glm3 , t_glm3 , 'sec' ) ,\n",
    "         ('CART', mse_cart , t_cart , 'sec' ) ,\n",
    "         ('BAG', mse_bag , t_bag , 'sec' ) ,\n",
    "         ('RF', mse_rf , t_rf , 'sec' ) ,\n",
    "         ('XGB', mse_xgb , t_xgb , 'sec' ) , \n",
    "         ('BART', mse_bart , t_bart , 'sec' ) ,\n",
    "         ('MARS', mse_mars , t_mars , 'sec' ) ,\n",
    "         ('RF_best', mse_rf_best , t_rf_best , 'sec' ) ,  \n",
    "         ('FNN', mse_fnn , t_fnn , 'sec' )]\n",
    "MSE = pd.DataFrame(MSE_data, columns=[\"Methode\", \"MSE\", \"Zeit\", \"Einheit\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2020-08-30T09:34:19.978015Z",
     "iopub.status.busy": "2020-08-30T09:34:19.972137Z",
     "iopub.status.idle": "2020-08-30T09:34:20.171925Z",
     "shell.execute_reply": "2020-08-30T09:34:20.171350Z"
    },
    "papermill": {
     "duration": 0.264945,
     "end_time": "2020-08-30T09:34:20.172077",
     "exception": false,
     "start_time": "2020-08-30T09:34:19.907132",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "x = MSE['Methode']\n",
    "y = pd.to_numeric(MSE['MSE'])\n",
    "\n",
    "diagramm = plt.figure(figsize=(10,6))\n",
    "plt.style.use('ggplot')\n",
    "\n",
    "plt.axhline(y=mse_rf_best, color = \"#348ABD\", label = \"mse_rf_best\")\n",
    "plt.axhline(y=mse_fnn, color = \"#988ED5\", label = \"mse_fnn\")\n",
    "            \n",
    "plt.plot(x, y, linestyle='-', marker='o', label = \"MSE\")\n",
    "plt.title('Übersicht der MSE')\n",
    "plt.ylabel('MSE')\n",
    "plt.xlabel('Verfahren')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "execution_count": null,
   "metadata": {
    "papermill": {
     "duration": 0.054791,
     "end_time": "2020-08-30T09:34:20.282954",
     "exception": false,
     "start_time": "2020-08-30T09:34:20.228163",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "<a id=\"Links\"></a>\n",
    "# Links\n",
    "Im folgenden geben wir eine kurze Zusammenfassung der im Text verwendeten Links an.\n",
    "* Alex Diana, Jim E. Griffin, Jaideep Oberoi and Aji Yao - Machine-Learning Methods for Insurance Applications - A Survey\n",
    "    - Studie: [https://www.soa.org/globalassets/assets/files/resources/research-report/2019/machine-learning-methods.pdf](https://www.soa.org/globalassets/assets/files/resources/research-report/2019/machine-learning-methods.pdf)\n",
    "    - Jupyter Notebook: [https://www.soa.org/globalassets/assets/files/resources/research-report/2019/group-ltc-jupyter-file.zip](https://www.soa.org/globalassets/assets/files/resources/research-report/2019/group-ltc-jupyter-file.zip)\n",
    "    - Datensatz: [https://www.soa.org/globalassets/assets/files/resources/research-report/2019/group-ltd-data-set.zip](https://www.soa.org/globalassets/assets/files/resources/research-report/2019/group-ltd-data-set.zip) \n",
    "\n",
    "* Mervyn Kopinsky - Predicting Group Long Term Disability Recovery and Mortality Rates using Tree Models\n",
    "    - Übersicht: [https://www.soa.org/resources/experience-studies/2017/2017-gltd-recovery-mortality-tree/](https://www.soa.org/resources/experience-studies/2017/2017-gltd-recovery-mortality-tree/)\n",
    "    - Studie: [https://www.soa.org/globalassets/assets/files/research/projects/2017-gltd-recovery-mortality-tree.pdf](https://www.soa.org/globalassets/assets/files/research/projects/2017-gltd-recovery-mortality-tree.pdf)\n",
    "\n",
    "* Roger Martin (Committee Chair), Barry Allen, Edd Bailey, Andy Baillargeon, Warren Cohen, Tom Corcoran, Fred Flores, Steve Garfield, Geoff Gerow, Scott Haglund, Paul Hitchcox, Rick Leavitt, Al Livingood, Ray Marra, Jacob McCoy, Eric Poirier, Ray Siwek, Ali Zaidi, Feng Zhao - 2008 Group Long Term Disability (Experience Table Report)\n",
    "    - Studie: [https://www.soa.org/globalassets/assets/Files/Research/Exp-Study/research-2008-ltd-gltd-table.pdf](https://www.soa.org/globalassets/assets/Files/Research/Exp-Study/research-2008-ltd-gltd-table.pdf)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  },
  "papermill": {
   "duration": 53.826793,
   "end_time": "2020-08-30T09:34:20.445347",
   "environment_variables": {},
   "exception": null,
   "input_path": "__notebook__.ipynb",
   "output_path": "__notebook__.ipynb",
   "parameters": {},
   "start_time": "2020-08-30T09:33:26.618554",
   "version": "2.1.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}