Comparative Evaluation Machine Learning and Deep Learning Models for Daily Air Quality Index Prediction

The prediction of air quality is essential for ensuring public health and safety, especially in urban areas with high pollution levels. The Air Quality Index (AQI) provides a standardized method for monitoring air pollution, incorporating various pollutants such as PM2.5, PM10, nitrogen dioxide (NO2), carbon monoxide (CO), and ozone (O3). Accurate forecasting of AQI levels helps to take timely actions to mitigate the health risks associated with poor air quality, such as respiratory diseases and cardiovascular conditions. Traditionally, statistical models and machine learning (ML) algorithms have been applied for AQI prediction, but with the advancement of deep learning (DL) models, there is an opportunity to improve prediction accuracy by handling complex patterns and temporal dependencies in air quality data.This paper investigates the performance of several machine learning and deep learning models for daily AQI prediction. The study includes traditional ML models like Linear Regression, Decision Trees, and Random Forest, as well as a state-of-the-art deep learning model—Long Short-Term Memory (LSTM) networks, which are particularly suitable for time-series forecasting. These models are evaluated based on their ability to predict AQI levels using historical air quality data, meteorological variables, and pollutant concentrations. The dataset used for this study is derived from real-time air quality monitoring stations and includes daily readings of various pollutants over several years. The models are trained and validated using standard performance metrics, including Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and R-squared (R²) values. Our findings reveal that while traditional ML models offer reasonable prediction accuracy, deep learning models, specifically LSTM networks, significantly outperform them. The LSTM model shows superior predictive accuracy due to its ability to capture long-term dependencies in the time-series data, making it highly effective for forecasting AQI. The results also demonstrate that LSTM outperforms Random Forest and Linear Regression in terms of both prediction accuracy and computational efficiency. offering a reliable tool for policymakers, environmentalists, and public health authorities to take preventive actions in case of deteriorating air quality.