Machine Learning based predictions for NFL running backs based on combine performance
I used combine data to make two separate projection model for NFL running backs' careers using pro-football-reference's career AV (added value) as the metric for success. Combine data turned out to be very limitedly connected to NFL success. On average, the linear model return a 0.07 correlation, while the neural net model returned about 0 with some fluctuations up and down depending on datasets.