RSTGCN: Railway-centric Spatio-Temporal Graph Convolutional Network for Train Delay Prediction
The episode analyzes 'RSTGCN,' a Spatio-Temporal Graph Convolutional Network designed to forecast average train delays across an entire railway network. Using a comprehensive operational dataset from Indian Railways, the authors discuss how the model integrates spatial and temporal data. The results show it outperforms existing models, providing tools for real-time risk assessment in dispatch systems.