Predicting stadium attendance for major league baseball
** Follow this link to watch the presentation of our solution: https://www.youtube.com/watch?v=o_j2os7tL1M **
- Identified factors driving stadium attendance and developed dynamic prediction model using ensemble and sequential modeling to forecast attendance with 8.9% MAPE, 28% lesser than baseline.
- SARIMAX model was used to address the causality between attendance and important features (game, player, geography and economy level indicators)
- For the prediction task, various models were tried including tree based and sequuential models.
- Two types of predictions were required - static to predict attendance of games ahead of the season start and dynamic to predict attendance while the season is underway
- Bayesian optimization technique was used to optimize game schedule for MLB 2023
