An Adaptive Adjacency Matrix-based Graph Convolutional Recurrent Network For Air Quality Prediction

A novel model called adaptive adjacency matrix-based graph convolutional recurrent network (AAMGCRN) is proposed for accurate air quality prediction. The model uses Point of Interest (POI) data and meteorological data to construct a self-ringing adjacency matrix and passes pollutant data through a Graph Convolutional Network (GCN) unit embedded in LSTM units to learn spatio-temporal dependencies. Experimental results show better predicted effects compared to other baseline models. This method can provide more accurate air quality predictions and support public health protection.