U.S. Air Quality Index Prediction with Regression Models

This project explored the use of geographic, pollutant, and population-based features to predict daily AQI values. After establishing a linear regression baseline (RMSE: 34.53), we tested Ridge Regression (RMSE: 34.42), which offered only slight improvement. However, Random Forest Regression achieved a significantly lower RMSE of 30.82, demonstrating its strength in capturing nonlinear relationships and feature interactions. These results showed a clear performance gap between linear and ensemble methods in modeling environmental data. The project highlighted how model selection directly affects predictive accuracy and laid the groundwork for future enhancements using weather and land-use variables.