DAILY MAXIMUM AQI PREDICTION IN TASHKENT USING CONVOLUTIONAL NEURAL NETWORKS (CNN)

• Abdurasul Bobonazarov

  Department of Software Engineering, Millat Umidi University

  Author

• Vazira Holboeva

  Department of Software Engineering, Millat Umidi University

  Author

Air pollution remains one of the most critical environmental challenges in urban regions. Accurate forecasting of air quality indices enables authorities and citizens to take preventive measures against harmful exposure. This study presents a deep learning-based approach for predicting the next-day maximum Air Quality Index (AQI) in Tashkent, Uzbekistan, using a Convolutional Neural Network (CNN). Three years of open-access air quality and meteorological data collected from the People’s Friendship Square monitoring station were aggregated at a daily level. The proposed multivariate CNN model leverages historical AQI, PM2.5 concentration, temperature, humidity, and precipitation values to forecast extreme pollution events. Experimental results demonstrate strong predictive performance, achieving an R² score of 0.793 and an RMSE of 43.22, indicating the model’s capability to capture temporal patterns and pollution peaks effectively.

• pdf

This work is licensed under a Creative Commons Attribution 4.0 International License.