Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

RSTGCN: Railway-centric Spatio-Temporal Graph Convolutional Network for Train Delay Prediction

Koyena Chowdhury, Paramita Koley, Abhijnan Chakraborty, Saptarshi Ghosh

TL;DR

RSTGCN tackles the challenge of predicting station-level average delays across India's vast railway network. It introduces a three-branch spatio-temporal GCN that fuses recent, daily, and weekly history, augmented by station-specific features (including hourly headway) and a train-frequency-aware spatial attention mechanism, with Chebyshev graph convolution and a final ReLU to reflect real-world delay behavior. The authors curate the first nationwide Indian Railway Network dataset (4735 stations, 9336 edges, 3892 trains across 17 zones) and demonstrate that RSTGCN consistently outperforms strong baselines (including TSTGCN) across 1–3 hour horizons and large zones, with robust long-horizon performance in the largest zone. The work delivers both methodological advances and a valuable public dataset to advance research in large-scale rail delay forecasting and operational planning.

Abstract

Accurate prediction of train delays is critical for efficient railway operations, enabling better scheduling and dispatching decisions. While earlier approaches have largely focused on forecasting the exact delays of individual trains, recent studies have begun exploring station-level delay prediction to support higher-level traffic management. In this paper, we propose the Railway-centric Spatio-Temporal Graph Convolutional Network (RSTGCN), designed to forecast average arrival delays of all the incoming trains at railway stations for a particular time period. Our approach incorporates several architectural innovations and novel feature integrations, including train frequency-aware spatial attention, which significantly enhances predictive performance. To support this effort, we curate and release a comprehensive dataset for the entire Indian Railway Network (IRN), spanning 4,735 stations across 17 zones - the largest and most diverse railway network studied to date. We conduct extensive experiments using multiple state-of-the-art baselines, demonstrating consistent improvements across standard metrics. Our work not only advances the modeling of average delay prediction in large-scale rail networks but also provides an open dataset to encourage further research in this critical domain.

RSTGCN: Railway-centric Spatio-Temporal Graph Convolutional Network for Train Delay Prediction

TL;DR

RSTGCN tackles the challenge of predicting station-level average delays across India's vast railway network. It introduces a three-branch spatio-temporal GCN that fuses recent, daily, and weekly history, augmented by station-specific features (including hourly headway) and a train-frequency-aware spatial attention mechanism, with Chebyshev graph convolution and a final ReLU to reflect real-world delay behavior. The authors curate the first nationwide Indian Railway Network dataset (4735 stations, 9336 edges, 3892 trains across 17 zones) and demonstrate that RSTGCN consistently outperforms strong baselines (including TSTGCN) across 1–3 hour horizons and large zones, with robust long-horizon performance in the largest zone. The work delivers both methodological advances and a valuable public dataset to advance research in large-scale rail delay forecasting and operational planning.

Abstract

Accurate prediction of train delays is critical for efficient railway operations, enabling better scheduling and dispatching decisions. While earlier approaches have largely focused on forecasting the exact delays of individual trains, recent studies have begun exploring station-level delay prediction to support higher-level traffic management. In this paper, we propose the Railway-centric Spatio-Temporal Graph Convolutional Network (RSTGCN), designed to forecast average arrival delays of all the incoming trains at railway stations for a particular time period. Our approach incorporates several architectural innovations and novel feature integrations, including train frequency-aware spatial attention, which significantly enhances predictive performance. To support this effort, we curate and release a comprehensive dataset for the entire Indian Railway Network (IRN), spanning 4,735 stations across 17 zones - the largest and most diverse railway network studied to date. We conduct extensive experiments using multiple state-of-the-art baselines, demonstrating consistent improvements across standard metrics. Our work not only advances the modeling of average delay prediction in large-scale rail networks but also provides an open dataset to encourage further research in this critical domain.

Paper Structure

This paper contains 17 sections, 7 equations, 2 figures, 9 tables.

Table of Contents

  1. Introduction
  2. Dataset and Problem statement
  3. Dataset Gathered and the Indian Railway Network
  4. Problem statement
  5. RSTGCN: Proposed framework
  6. Station-specific features
  7. Model details
  8. Experimental Evaluation
  9. Data processing
  10. Metrics
  11. Baselines
  12. Experimental setup
  13. Delay prediction results
  14. Long-horizon forecasting
  15. Ablation Study
  16. ...and 2 more sections

Figures (2)

  • Figure 1: RSTGCN Model Architechture
  • Figure 2: Delay prediction performance of RSTGCN with $2$ best baselines on the largest zone (SCR) for long-term prediction (up to $12$ hours) for all the metrics.