Air-Out-Probabilities

This project develops machine learning models to predict air-out probabilities for batted balls in Minor League Baseball using 2023 season data. The dataset includes play metadata, Trackman metrics, and player defensive data. The project involves:

1. Data Preprocessing: Handling missing values, feature selection, and encoding categorical variables.
2. Modeling: Comparing Logistic Regression and Random Forest models for prediction accuracy. The Random Forest model achieved superior performance with a validation log loss of 0.283.
3. Insights and Applications: Results inform player development and coaching strategies by providing actionable metrics on defensive performance.