The project uses classification models with double cross-validation to predict if a player would be a starter, Pro Bowl, or All-Pro player. The positions of quarterbacks, running backs, and wide receivers were used for the predictions. Another objective was to see if college statistics or combine stats were a better predictor.