Short-term Prediction of Vessel Traffic Flow Based on Enhanced Spatiotemporal Transformer – Reproducible Code & Processed Qiongzhou Strait AIS Flow Datasets

This archive contains all reproducible materials for the manuscript *Short-term Prediction of Vessel Traffic Flow Based on Enhanced Spatiotemporal Transformer*. File contents overview: 1. Full source code of the proposed ESTformer model and four mainstream baseline models (STAEformer, FSTformer, AGCRN, Mamba), including training, validation, ablation test and metric calculation scripts. 2. Processed hourly vessel traffic flow matrix derived from 3-month Qiongzhou Strait AIS raw records, which serves as the unique input data of all prediction experiments. 3. Environment configuration file, one-click experiment running script, dataset specification document and MIT open-source license.