Appex 17 – Decision Trees

STA 363 - Spring 2023

Set up

Login to RStudio Pro

• Note: if you are off campus, you will need to use a VPN to connect
• Go to rstudio.deac.wfu.edu

Step 1: Create a New Project

Click File > New Project

Step 2: Click “Version Control”

Click the third option.

Step 3: Click Git

Click the first option

Step 4: Copy my starter files

Paste this link in the top box (Repository url):

https://github.com/sta-363-s23/appex-20.git

Part 1

1. Pull in the application exercise files from:

https://github.com/sta-363-s23/appex-20.git

2. Run install.packages("keras") once in the console
3. Run keras::install_keras() once in the console
4. When it asks if you want to install Miniconda type Y

Part 2

1. Create a recipe that uses all of the included variables to predict Salary
2. Remove any missing data for the outcome, impute data for the remaining predictors (HINT: in step_naomit() set skip = FALSE to make sure it does this when we are prepping the data)
3. Make all nominal variables dummy variables
4. Normalize all predictors using step_normalize()

Part 3

1. Using the Hitters data, split into 2/3 training and 1/3 testing data
2. Create datasets for the training and testing data
3. Run the recipe on the training and testing data
4. Extract just the predictors into a data frame and then convert this into a matrix for both the training and testing pre-processed data

Part 4

1. Fit a model with 40 activations in a hidden layer with 50% dropout and a batch size of 30
2. Run your model for 1,000 iterations (epochs)
3. Use your testing data as validation
4. Plot the output