Spatiotemporal Forecasting for Regional Express Pickup Demand

To improve the efficiency of the parcel pickup process, accurately identifying areas with high demand has become a pressing challenge. Using data of Points of Interest, the hierarchical graph infomax (HGI) method is employed to generate regional embeddings. These embeddings, along with historical pickup demand series, are then input into a graph multi-head attention network (GMAN) for prediction, forming the proposed HGI-GMAN model. Experimental results from a real-world dataset provided by an express company indicate that the HGI-GMAN model outperforms five classical baseline models across various regression (RMSE, R²) and classification (macro-F1) metrics. Additionally, sensitivity analyses and ablation studies confirm the model’s robustness and the effectiveness of its features.