✈️ NBA Schedule Optimization

Developed an optimization framework combining integer linear programming and machine learning to redesign single-matchup NBA schedules.​ 

 

Impact: Estimated $2.5M in annual travel cost savings and a 45% reduction in carbon emissions relative to baseline schedules. ​

Stack: Python, PuLP, scikit-learn, PyTorch, Seaborn, Tableau

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

 

​​

​​

​

​

​

​​

​

​

📊 NBA Clutch Player of the Year Prediction Model

Built regression-based models on historical NBA box score data (1996–present) to construct a novel, data-driven clutch performance metric.

​

Impact: Reached ~90% alignment with 10 major betting platforms and correctly predicted Stephen Curry as the 2023–24 Clutch Player of the Year. Deployed an interactive R Shiny application enabling filtering of candidates across 16 customizable parameters.

​

Stack: R, R Shiny

​

​

​

​

​

​

​​

​

​

​

​

​

​

​

​

​

​

​

​​

​

🏀 NBA Pass-or-Shoot Image Classification System

Designed and trained a convolutional neural network to classify NBA in-game images as pass or shoot actions using labeled broadcast frames.

Impact: Achieved 96.25% classification accuracy on ~2,000 images. Developed a Streamlit interface for real-time drag-and-drop inference, with integration into Synergy Sports game feeds to predict live decision-making insights.

​

Stack: Python, TensorFlow, BeautifulSoup, Requests, Streamlit

​​​​​​​

​​

​​​​​​

🎯 NBA Shot Location Efficiency Analysis

Analyzed league-wide shot location, frequency, and efficiency data since 2000 to identify spatial trends in scoring effectiveness.

​

Impact: Identified the region just inside the corner three as the most accurate shot zone across seasons. Results were visualized through Tableau heatmaps and interactive dashboards to support exploratory and comparative analysis.

​

Stack:  R, Python, Tableau