Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

WEST GCN-LSTM: Weighted Stacked Spatio-Temporal Graph Neural Networks for Regional Traffic Forecasting

Theodoros Theodoropoulos, Angelos-Christos Maroudis, Antonios Makris, Konstantinos Tserpes

TL;DR

The paper addresses regional traffic forecasting by extending spatio-temporal graph neural networks with a weighted stacked GCN encoder (WEST) and an LSTM decoder, enabling multi-hop spatial aggregation and temporal modeling. Two policies—Shared Borders and Adjustable Hops—guide the construction of the weighted adjacency and the number of GCN layers to adapt to regional topology and population speeds, respectively. Across Berlin and Central Park datasets (SUMO-based, with 6 regions), the WEST GCN-LSTM and its variant WE GCN-LSTM substantially outperform 19 baselines, with consistent gains in RMSE, MSE, and MAE, and ablation confirms each component’s contribution. The approach demonstrates robust regional forecasting by fusing spatial topology, population dynamics, and temporal patterns, offering a scalable framework for IoE-enabled urban mobility analytics.

Abstract

Regional traffic forecasting is a critical challenge in urban mobility, with applications to various fields such as the Internet of Everything. In recent years, spatio-temporal graph neural networks have achieved state-of-the-art results in the context of numerous traffic forecasting challenges. This work aims at expanding upon the conventional spatio-temporal graph neural network architectures in a manner that may facilitate the inclusion of information regarding the examined regions, as well as the populations that traverse them, in order to establish a more efficient prediction model. The end-product of this scientific endeavour is a novel spatio-temporal graph neural network architecture that is referred to as WEST (WEighted STacked) GCN-LSTM. Furthermore, the inclusion of the aforementioned information is conducted via the use of two novel dedicated algorithms that are referred to as the Shared Borders Policy and the Adjustable Hops Policy. Through information fusion and distillation, the proposed solution manages to significantly outperform its competitors in the frame of an experimental evaluation that consists of 19 forecasting models, across several datasets. Finally, an additional ablation study determined that each of the components of the proposed solution contributes towards enhancing its overall performance.

WEST GCN-LSTM: Weighted Stacked Spatio-Temporal Graph Neural Networks for Regional Traffic Forecasting

TL;DR

The paper addresses regional traffic forecasting by extending spatio-temporal graph neural networks with a weighted stacked GCN encoder (WEST) and an LSTM decoder, enabling multi-hop spatial aggregation and temporal modeling. Two policies—Shared Borders and Adjustable Hops—guide the construction of the weighted adjacency and the number of GCN layers to adapt to regional topology and population speeds, respectively. Across Berlin and Central Park datasets (SUMO-based, with 6 regions), the WEST GCN-LSTM and its variant WE GCN-LSTM substantially outperform 19 baselines, with consistent gains in RMSE, MSE, and MAE, and ablation confirms each component’s contribution. The approach demonstrates robust regional forecasting by fusing spatial topology, population dynamics, and temporal patterns, offering a scalable framework for IoE-enabled urban mobility analytics.

Abstract

Regional traffic forecasting is a critical challenge in urban mobility, with applications to various fields such as the Internet of Everything. In recent years, spatio-temporal graph neural networks have achieved state-of-the-art results in the context of numerous traffic forecasting challenges. This work aims at expanding upon the conventional spatio-temporal graph neural network architectures in a manner that may facilitate the inclusion of information regarding the examined regions, as well as the populations that traverse them, in order to establish a more efficient prediction model. The end-product of this scientific endeavour is a novel spatio-temporal graph neural network architecture that is referred to as WEST (WEighted STacked) GCN-LSTM. Furthermore, the inclusion of the aforementioned information is conducted via the use of two novel dedicated algorithms that are referred to as the Shared Borders Policy and the Adjustable Hops Policy. Through information fusion and distillation, the proposed solution manages to significantly outperform its competitors in the frame of an experimental evaluation that consists of 19 forecasting models, across several datasets. Finally, an additional ablation study determined that each of the components of the proposed solution contributes towards enhancing its overall performance.
Paper Structure (16 sections, 10 equations, 4 figures, 3 tables, 3 algorithms)

This paper contains 16 sections, 10 equations, 4 figures, 3 tables, 3 algorithms.

Table of Contents

  1. Introduction
  2. Literature Review
  3. Problem Formulation
  4. Proposed Solution
  5. WEST GCN-LSTM
  6. WEighted STacked (WEST) Graph Convolutional Networks (GCN)
  7. Long Short Term Memory (LSTM)
  8. WEST GCN-LSTM
  9. Shared Borders Policy
  10. Adjustable Hops Policy
  11. Experimental Evaluation
  12. Evaluation Metrics
  13. Benchmark Datasets
  14. Experimental Results
  15. Discussion
  16. ...and 1 more sections

Figures (4)

  • Figure 1: GCN-LSTM architecture.
  • Figure 2: An overview of the proposed solution.
  • Figure 3: Densities of the regional traffic volumes that correspond to the: (1) Berlin , (2) Central Park, (3) Central Park (Low), and (4) Central Park (High) datasets.
  • Figure 4: Normalized RMSE values that correspond to the proposed solution (ours), to GCN-LSTM (binary), as well as to the best LR, ML, ED, and GCN-LSTM solutions.