AQI Prediction Project

Overview

This project analyzes and models Air Quality Index (AQI) data across 24 U.S. states using historical data from 2023 to predict AQI in 2024. We explore both linear and non-linear machine learning models to understand pollutant and weather impacts on air quality.

Goals

• Conduct a comparative analysis to test the hypothesis that both urban and rural areas experience AQI issues due to different pollution sources.
• Predict 2024 AQI using 2023 state-level averages.
• Compare model performance: Baseline, Linear Regression, and Random Forest.
• Extract insights using feature importance from machine learning models.
• Extend the model to daily, county-level AQI data and classify AQI into categories (e.g., Good, Moderate, Unhealthy).

Data

• Source: U.S. EPA AQI and meteorological datasets
• Files Used:
  • 24StateAQI_2023.csv
  • 24StateAQI_2024.csv
  • daily_aqi_by_county_2023.csv
  • daily_aqi_by_county_2024.csv

How to Run

1. Load and merge datasets (unzip first)
2. Preprocess and aggregate AQI by state.
3. Train and evaluate models.
4. Visualize results and feature importance.

Requirements

• Python 3.8+
• pandas, scikit-learn, matplotlib, seaborn, numpy

---