Every season, the NBA's best young talent competes for the Rookie of the Year (ROY) award. While raw scoring ability often catches the public's eye, the award is about more than just points. Team success, efficiency, and overall impact all play a role. But can machine learning help us predict the next winner?
In this project, I built a model to forecast the 2025 NBA Rookie of the Year, using historical ROY data and advanced analytics. By analyzing past winners and their performance metrics, my goal was to develop a data-driven approach to predicting the league's top rookie.
Traditionally, voters consider factors such as:
Using these factors, I gathered rookie performance data and created a machine learning model to predict future ROY winners.
To build the model, I compiled data from past NBA Rookie of the Year winners, focusing on key statistics that correlated with past selections. After performing exploratory data analysis (EDA), I identified the most important features:
These variables formed the basis for my predictive model.
After testing multiple models, Random Forest Regression provided the best performance.
This model outperformed others in predicting past ROY winners and was ultimately chosen to forecast the 2025 ROY race.
Applying my trained model to the 2025 NBA rookie class, the projected Rookie of the Year is:
This project demonstrates how machine learning can enhance sports analysis and prediction. While no model is perfect, data-driven insights provide a structured way to evaluate the ROY race. The model successfully identified past winners and projects Stephon Castle as the 2025 Rookie of the Year.
🚀 Interested in the full analysis? View the project on GitHub.