This project was done using RStudio. A variety of packages such as ggplot2, gganimate, dplyr and ROCR were installed in order to construct our analysis and visuals.
|
Logistic Regression |
Why Logistic Regression |
Logistic Regression is a regression algorithm that outputs the odds that a shot is made given a matrix of input factors. The user then has to determine the cutoff probability to result in a hard classification, yes or no.
|
Logistic Regression is very interpretable as each of the Betas in the model can be read as the increase in the odds of making the shots. This kind of analysis can help current players understand where they are significantly better at making a basket.
|
Data Preparation |
The dependent variable is the probability of making the shot which is a number from 0-1. The regression has information about the shot type (in binary form) and game fixed effects and season fixed effects. The way to interpret each coefficient is 'an increase of 'X' by one would increase the odds of making the shot by e^(Beta).
|