Presenter(s)

    Jonah Mergler

    Comments

    9:00-10:15, Kennedy Union Ballroom

    Download

    Download Project (188 KB)

    Description

    Growing up in a family of devoted Flyers fans, I developed a deep appreciation for basketball, especially after watching Obi Toppin. The intensity, strategic plays, and constant innovation of the game fueled my passion. When choosing my capstone project, incorporating basketball was an obvious choice. Since college basketball was set to end before my presentation at Stander Symposium, I decided to focus on the NBA, where data is more accessible and player rosters remain stable for longer periods.This project utilizes machine learning and data analytics to predict the 2025 NBA champion by analyzing the last 20 years of team statistics alongside this season’s data. I compiled a dataset featuring key performance metrics, including wins, losses, field goal percentages, rebounds (offensive, defensive, and total), and steals, both for and against teams. Using this data, I developed predictive models to assess each team's likelihood of winning the championship.To classify potential champions, I employed supervised learning techniques such as logistic regressions, LDA, QDA, KNN, random forests, and gradient boosting. The model was trained and validated using historical NBA data spanning two decades. Additionally, I applied various feature selection techniques, including forward selection and LASSO, alongside in-depth exploratory data analysis (EDA) to determine the most significant predictors of championship success.This project aims to offer fans, analysts, and sports bettors a data-driven approach to forecasting the NBA champion. My findings highlight the power of machine learning in sports prediction, demonstrating how data analytics can uncover patterns that may not be immediately apparent through traditional analysis.

    Publication Date

    4-23-2025

    Project Designation

    Capstone Project

    Primary Advisor

    Gayan J. Warahena Liyanage

    Primary Advisor's Department

    Mathematics

    Keywords

    Stander Symposium, College of Arts and Sciences

    Institutional Learning Goals

    Practical Wisdom

    Numbers Don't Lie: Forecasting the NBA Champion with Machine Learning

    Share

    COinS