NBA Shot Predictor

Two scikit-learn models trained on public NBA datasets. Originally a 2019 BYU class project that called Azure ML Studio (retired by Microsoft in 2024); modernized in 2026 with locally-trained random forests.

Model 1

Shot Outcome

Given pre-shot tracking features (shot clock, distance, defender, dribbles), predict the probability the shot goes in.

Trained on 128k shots from the 2014-15 season. AUC 0.634.

Model 2

Salary Estimate

Given a player's season stats (age, games, minutes, win shares), predict what their salary should be.

Trained on 12.7k player-seasons from 1990-2017. Median error ~$940k.