About

This project analyzes the operational drivers behind the Boston Celtics' win probability using an 82-game performance dataset on Zerve. The analysis follows a complete data science pipeline: cleaning raw game logs, engineering feature ratios for 3-point volume and shooting efficiency, conducting visual exploratory data analysis, and fitting a Logistic Regression model to identify key winning indicators.

The exploratory analysis revealed a sharp efficiency threshold—when Boston shoots above 37.5% from beyond the arc while maintaining 11 or fewer turnovers, their win probability reaches over 85%. Logistic regression modeling confirmed that 3PT percentage and field goal efficiency are the primary positive drivers of point margin, while turnover rate is the largest negative factor. The final output combines interactive visualizations and feature importance rankings to provide clear, actionable insights into game execution strategy.