This project proposes an Attention-Enhanced Spatio-Temporal Graph Convolutional Network (AE-STGCN) for air quality prediction in rapidly industrializing regions. The model fuses heterogeneous data sources—air pollutant measurements, meteorological variables, and point-of-interest (POI) data—into a grid-based graph representation. Spatial–temporal attention highlights the most important regions and time steps, improving predictive accuracy and interpretability.