Including a categorical variable in Notebook #4 when finding R^2..

KHAFFERTY

I am not familiar with what is happening when students include a categorical in their model/equation, so I told them to stick to quantitative variables. Was that the wrong advice if they want to submit to the national competition?

wcraine

I can't answer the question about whether or not the students will be disadvantaged in the competition by not using categorical variables, but I can explain what R is doing with categorical variables in multiple regression. I think...

When you add a categorical variable to your model, it converts the levels of that variable to variables in their own right, coded as binary, 0 means doesn't fit that category, 1 means fits that category. Then it adds those variables to the regression as quantitative variables.

For example, if you chose to include hbcu (historically Black colleges and universities), R would create variables hbcuNo and hbcuYes. Morehouse College would have 0 added to its case under hbcuNo and 1 added to its case under hbcuYes, while Harvard would have 1 added to its case under hbcuNo and 0 added to its case under hbcuYes.

This, of course, would get wild if you tried to include the college name as a predictor!

You can see what it does by adding a categorical variables to my_model then running summary(my_model) to have a look.

ivonnem10

Great Question! The AP Challenge exercises mainly focus on regression methods but there are still a lot of things that can be done to get a great model. That being said, there are the main things that we look at for students submissions:

Model seed should be set at 2024
Their model should be trained on the train data (my model <- lm(default_rate ~ …., data = train)
Predictions are being made on the test dataset

As long as they follow these guidelines their submission should be valid. Now for what variables they decide to include and how to include them, as long as they are following the above their model would be considered for the competition.

To get a better idea of how competitive the submissions are I would suggest you read last year's competition article (https://medium.com/data-science-4-everyone/after-the-ap-data-science-champions-announced-fd405bcdb883)