diff --git a/GP_Solutions.qmd b/GP_Solutions.qmd
index 6be9246..7eda00a 100644
--- a/GP_Solutions.qmd
+++ b/GP_Solutions.qmd
@@ -1,10 +1,23 @@
---
-title: "GP_Solutions"
-format: html
-editor: visual
+title: "VectorByte Methods Training: Introduction to Gaussian Processes for Time Dependent Data (Practical - solution)"
+author:
+ - name: Parul Patil
+ affiliation: Virginia Tech and VectorByte
+citation: true
+date: 2024-07-24
+date-format: long
+format:
+ html:
+ toc: true
+ toc-location: left
+ html-math-method: katex
+ css: styles.css
+bibliography: references.bib
+link-citations: TRUE
---
-##### Libraries
+
+# Libraries
```{r, warning=FALSE, message=FALSE, warn.conflicts = FALSE}
library(mvtnorm)
@@ -13,7 +26,7 @@ library(hetGP)
library(ggplot2)
```
-##### HetGP (sin wave eg)
+# HetGP (sin wave eg)
```{r}
# Your turn
@@ -50,7 +63,7 @@ nugs <- het_pred$nugs
summary(nugs)
```
-##### Challenges
+# Challenges
We need to load the data and the functions
@@ -86,7 +99,9 @@ fx.sin <- function(datetime, f1 = fx.iso_week){
}
```
-###### Fit a GP Model for the location "SERC" i.e. `site_number = 7`.
+## Fit a GP Model for the location "SERC"
+
+i.e. `site_number = 7`.
Just change site = 7
@@ -191,7 +206,7 @@ rmse <- sqrt(mean((yt_true - yt_pred)^2))
rmse
```
-###### Use an environmental predictor in your model. Following is a function `fx.green` that creates the variable given the `datetime` and the `location`.
+## Use an environmental predictor in your model. Following is a function `fx.green` that creates the variable given the `datetime` and the `location`.
Here is a snippet of the supporting file that you will use; You can look into the data.frame and try to plot `ker` for one site at a time and see what it yields.
@@ -215,7 +230,7 @@ fx.green <- function(datetime, site, site_info = df_green){
- set up X3 using `fx_green`
- Scale X3
-Setting up the target dataframe
+### Setting up the target dataframe
```{r}
@@ -234,7 +249,7 @@ df_train <- subset(df, df$datetime <= cutoff)
df_test <- subset(df, df$datetime > cutoff)
```
-Adding Greenness
+### Adding Greenness
```{r}
# Choose location
@@ -333,7 +348,7 @@ rmse <- sqrt(mean((yt_true - yt_pred)^2))
rmse
```
-###### Fit a GP Model for all the locations (*More advanced*).
+## Fit a GP Model for all the locations (*More advanced*).
```{r}
# GP function. This can be varied but easiest way is to just take in X, y, XX and return the predicted means and bounds.
diff --git a/docs/GP_Practical_files/figure-html/unnamed-chunk-6-1.png b/docs/GP_Practical_files/figure-html/unnamed-chunk-6-1.png
index 1b5f8dd..61f2972 100644
Binary files a/docs/GP_Practical_files/figure-html/unnamed-chunk-6-1.png and b/docs/GP_Practical_files/figure-html/unnamed-chunk-6-1.png differ
diff --git a/docs/GP_Solutions.html b/docs/GP_Solutions.html
index 747560b..53c343a 100644
--- a/docs/GP_Solutions.html
+++ b/docs/GP_Solutions.html
@@ -6,8 +6,10 @@
+
+
-VectorByte Training 2024 - GP_Solutions
+VectorByte Training 2024 - VectorByte Methods Training: Introduction to Gaussian Processes for Time Dependent Data (Practical - solution)