RandomForest.Rmd

---
title: "RandomForest"
author: "Bijan GURUNG"
date: "3/6/2022"
output:
  ioslides_presentation: default
  powerpoint_presentation: default
---

```{r}
```

## Outline

- A brief overview of RandomForest library in r
  - What is RandomForest?
  - Where do we use RandomForest?
  - How RandomForest is used in regression or classification?

## RandomForest

- To predict the result 
- Classify the result of data
- Classify a variable such as True or False (e.g. it rains/true or does not rain/false)
- Categorical variable or non-numeric variable
- The algorithm makes many decision trees (~forests) by random sampling or randomization (~random)
- RandomForest 
- Classification or Regression Analysis

## Slide with R Output

Install the required packages for dataset and analysis.

```{r}
#install.packages("stats")
#install.packages("dplyr")
#install.packages("randomForest")
```

- Uploading the required libraries
- We are using the data contained in library "stats"
```{r, message=FALSE}
library(stats)
library(dplyr)
library(randomForest)
```

## Working with "iris" dataset 
- data about measurement of plants
```{r}
mydata = iris 

head(mydata, 10)
```
- characteristics of the dataset

```{r}
str(mydata)
```

## Analysis using RandomForest
- Divide the entire dataset into 
  - Training dataset (set a model from the data)
  - Testing dataset (to test the model)
```{r}
index = sample(2, nrow(mydata), replace=TRUE, prob=c(0.7, 0.3))

Training = mydata[index == 1, ]

Testing = mydata[index == 2, ]
```

## randomForest function
- Treeline structure (branches and nodes) in the analysis to create a model
```{r}
RFM = randomForest(Species~., data = Training)
```

- Testing of the model on test dataset
```{r}
Species_Pred = predict(RFM, Testing)
```

## Result of the testing of the model into test dataset
- checking of the model in test data (cross validation)
```{r}
Testing$Species_Pred = Species_Pred

head(Testing, 10)
```

## Model accuracy
- High accuracy indicates a good model 
- Low accuracy indicates the model failed to capture all characteristics of the data 
- Checking of parameters in the analysis (optimization)
```{r}
CFM = table(Testing$Species, Testing$Species_Pred)
```

```{r}
Classification_Accuracy = sum(diag(CFM)/sum(CFM))
Classification_Accuracy
```

## 

- Application in other fields (landuse change, finance)
- Sources: https://www.youtube.com/watch?v=acFviblzijU, Easy ML 
- http://rpubs.com/BijanGURUNG/874775