Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

DST-TransitNet: A Dynamic Spatio-Temporal Deep Learning Model for Scalable and Efficient Network-Wide Prediction of Station-Level Transit Ridership

Jiahao Wang, Amer Shalaby

TL;DR

DST-TransitNet is introduced, a hybrid DL model for system-wide station-level ridership prediction that uses graph neural networks (GNN) and recurrent neural networks (RNN) to dynamically integrate the changing temporal and spatial correlations within the stations.

Abstract

Accurate prediction of public transit ridership is vital for efficient planning and management of transit in rapidly growing urban areas in Canada. Unexpected increases in passengers can cause overcrowded vehicles, longer boarding times, and service disruptions. Traditional time series models like ARIMA and SARIMA face limitations, particularly in short-term predictions and integration of spatial and temporal features. These models struggle with the dynamic nature of ridership patterns and often ignore spatial correlations between nearby stops. Deep Learning (DL) models present a promising alternative, demonstrating superior performance in short-term prediction tasks by effectively capturing both spatial and temporal features. However, challenges such as dynamic spatial feature extraction, balancing accuracy with computational efficiency, and ensuring scalability remain. This paper introduces DST-TransitNet, a hybrid DL model for system-wide station-level ridership prediction. This proposed model uses graph neural networks (GNN) and recurrent neural networks (RNN) to dynamically integrate the changing temporal and spatial correlations within the stations. The model also employs a precise time series decomposition framework to enhance accuracy and interpretability. Tested on Bogota's BRT system data, with three distinct social scenarios, DST-TransitNet outperformed state-of-the-art models in precision, efficiency and robustness. Meanwhile, it maintains stability over long prediction intervals, demonstrating practical applicability.

DST-TransitNet: A Dynamic Spatio-Temporal Deep Learning Model for Scalable and Efficient Network-Wide Prediction of Station-Level Transit Ridership

TL;DR

DST-TransitNet is introduced, a hybrid DL model for system-wide station-level ridership prediction that uses graph neural networks (GNN) and recurrent neural networks (RNN) to dynamically integrate the changing temporal and spatial correlations within the stations.

Abstract

Accurate prediction of public transit ridership is vital for efficient planning and management of transit in rapidly growing urban areas in Canada. Unexpected increases in passengers can cause overcrowded vehicles, longer boarding times, and service disruptions. Traditional time series models like ARIMA and SARIMA face limitations, particularly in short-term predictions and integration of spatial and temporal features. These models struggle with the dynamic nature of ridership patterns and often ignore spatial correlations between nearby stops. Deep Learning (DL) models present a promising alternative, demonstrating superior performance in short-term prediction tasks by effectively capturing both spatial and temporal features. However, challenges such as dynamic spatial feature extraction, balancing accuracy with computational efficiency, and ensuring scalability remain. This paper introduces DST-TransitNet, a hybrid DL model for system-wide station-level ridership prediction. This proposed model uses graph neural networks (GNN) and recurrent neural networks (RNN) to dynamically integrate the changing temporal and spatial correlations within the stations. The model also employs a precise time series decomposition framework to enhance accuracy and interpretability. Tested on Bogota's BRT system data, with three distinct social scenarios, DST-TransitNet outperformed state-of-the-art models in precision, efficiency and robustness. Meanwhile, it maintains stability over long prediction intervals, demonstrating practical applicability.

Paper Structure

This paper contains 20 sections, 15 equations, 23 figures, 3 tables.

Table of Contents

  1. INTRODUCTION
  2. Background
  3. Methodology
  4. Basic Building Blocks
  5. Gated Recurrent Unit (GRU)
  6. Graph Convolutional Networks
  7. Graph Attention Network
  8. DST-TransitNets
  9. Temporal Decomposition
  10. Dynamic Spatial Weight Calculation
  11. Spatio-Temporal Aggregation
  12. Long-term Prediction Using Short-term Prediction Model
  13. Pipeline for Real-World Deployment
  14. Experiments and Result Analysis
  15. Datasets
  16. ...and 5 more sections

Figures (23)

  • Figure 1: Unit structure of a GRU, showing how previous hidden states and current inputs are combined to produce the output at each timestep.
  • Figure 2: Message passing through a graph-structured dataset via GCN layers
  • Figure 3: Graph Attention Network (GAT) unit, demonstrating how attention mechanisms are used to weight the contributions of neighboring nodes dynamically.
  • Figure 4: Dynamic Spatio-Temporal TransitNet (DST-TransitNet) Model Structure
  • Figure 5: The architecture of the Spatio-Temporal Aggregation layer in DST-TransitNet, combining Dynamic Spatial Aggregation (via k-GNN and GATConv) and Temporal Aggregation (via stacked GRUs).
  • ...and 18 more figures