Machine Learning-Based Prediction of PM2.5 and PM10 Levels in Raipur’s Air Quality

Machine Learning-Based Prediction of PM2.5 and PM10 Levels in Raipur’s Air Quality

Dr. Shweta Sharma

Assistant Professor, Department of Geography

Mahant Laxminarayan Das college, Raipur, Chhattisgarh, India

Dr. Prem Kumar Chandrakar
Assistant Professor, Department of Computer Science
Mahant Laxminarayan Das college, Raipur, Chhattisgarh, India

Abstract:

Urban air quality is becoming an increasingly critical issue in India due to rapid urban and industrial growth. Raipur, the capital of Chhattisgarh, faces worsening air quality driven by transportation, construction, and industrial emissions. This paper investigates PM2.5 and PM10 pollutant trends and applies machine learning techniques to forecast short-term air quality using meteorological and environmental inputs. Historical air quality data spanning January 2018 to March 2024 was sourced from the Central Pollution Control Board (CPCB) and OpenAQ. After rigorous preprocessing, machine learning models such as Linear Regression, Random Forest, XGBoost, and Long Short-Term Memory (LSTM) were implemented and evaluated. Using an 80:20 training-test split and evaluation metrics including RMSE, MAE, and R², results showed LSTM and XGBoost provided the most accurate forecasts. These findings reinforce the effectiveness of machine learning in air quality prediction and support data-driven planning and policy-making for environmental management in Raipur.

Keywords: Air Quality, PM2.5, PM10, Raipur, Machine Learning, Regression Models, XGBoost, LSTM