A systematic study on PM2.5 and PM10 concentration prediction in air pollution using machine learning and deep learning model
Pranshu Patel, Swara Patel, Kanish Shah, Kedar Desai, Samrat Patel, Manan Shah, Samir Patel
Abstract
High concentrations of PM10 and PM2.5 in the air pose significant risks to human health, making accurate forecasting crucial for air quality management and public safety. This study evaluates predictive models for particulate matter concentrations using air quality monitoring data from Maharashtra state (2019–2023). To determine their predictive effectiveness, various machine learning models, including Linear Regression, Decision Tree, Random Forest, and XGBoost, were compared against a deep learning-based Long Short-Term Memory (LSTM) model. To quantify their accuracy, the models were assessed using R 2 Score and Root Mean Square Error (RMSE). Our results demonstrate that LSTM significantly outperformed traditional machine learning models, achieving R 2 scores ranging from 0.99 to 0.998 across the five cities analyzed. In contrast, other models, including Random Forest, Decision Tree, and XGBoost, exhibited R 2 scores between 0.15 and 0.96, indicating lower predictive accuracy. The findings highlight that LSTM models can effectively capture complex temporal dependencies in air quality data, making them a more reliable approach for forecasting PM2.5 and PM10 concentrations. These results underscore the potential of deep learning in enhancing air quality prediction models and aiding environmental decision-making.