Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Multiple Areal Feature Aware Transportation Demand Prediction

Sumin Han, Jisun An, Youngjun Park, Suji Kim, Kitae Jang, Dongman Lee

TL;DR

The paper tackles short-term transportation demand prediction by integrating polymorphic urban features through a multi-feature, sentinel-attention graph architecture. It introduces ST-MFGCRN, which splits temporal inputs into closeness, period, and trend components, uses spatio-temporal embedding, and learns multiple area-similarity graphs via a sentinel-augmented GCGRU for robust multi-feature fusion. Key contributions include sentinel attention to enable partial feature usage, a multi-graph GRU to fuse diverse areal signals, and a weighted fusion mechanism that adaptively combines information from different time units. Experiments on BusDJ and TaxiBJ show consistent gains over state-of-the-art baselines, demonstrating that incorporating rich areal features (land use, POI, demographics, and road data) improves predictive accuracy in real-world urban networks.

Abstract

A reliable short-term transportation demand prediction supports the authorities in improving the capability of systems by optimizing schedules, adjusting fleet sizes, and generating new transit networks. A handful of research efforts incorporate one or a few areal features while learning spatio-temporal correlation, to capture similar demand patterns between similar areas. However, urban characteristics are polymorphic, and they need to be understood by multiple areal features such as land use, sociodemographics, and place-of-interest (POI) distribution. In this paper, we propose a novel spatio-temporal multi-feature-aware graph convolutional recurrent network (ST-MFGCRN) that fuses multiple areal features during spatio-temproal understanding. Inside ST-MFGCRN, we devise sentinel attention to calculate the areal similarity matrix by allowing each area to take partial attention if the feature is not useful. We evaluate the proposed model on two real-world transportation datasets, one with our constructed BusDJ dataset and one with benchmark TaxiBJ. Results show that our model outperforms the state-of-the-art baselines up to 7\% on BusDJ and 8\% on TaxiBJ dataset.

Multiple Areal Feature Aware Transportation Demand Prediction

TL;DR

The paper tackles short-term transportation demand prediction by integrating polymorphic urban features through a multi-feature, sentinel-attention graph architecture. It introduces ST-MFGCRN, which splits temporal inputs into closeness, period, and trend components, uses spatio-temporal embedding, and learns multiple area-similarity graphs via a sentinel-augmented GCGRU for robust multi-feature fusion. Key contributions include sentinel attention to enable partial feature usage, a multi-graph GRU to fuse diverse areal signals, and a weighted fusion mechanism that adaptively combines information from different time units. Experiments on BusDJ and TaxiBJ show consistent gains over state-of-the-art baselines, demonstrating that incorporating rich areal features (land use, POI, demographics, and road data) improves predictive accuracy in real-world urban networks.

Abstract

A reliable short-term transportation demand prediction supports the authorities in improving the capability of systems by optimizing schedules, adjusting fleet sizes, and generating new transit networks. A handful of research efforts incorporate one or a few areal features while learning spatio-temporal correlation, to capture similar demand patterns between similar areas. However, urban characteristics are polymorphic, and they need to be understood by multiple areal features such as land use, sociodemographics, and place-of-interest (POI) distribution. In this paper, we propose a novel spatio-temporal multi-feature-aware graph convolutional recurrent network (ST-MFGCRN) that fuses multiple areal features during spatio-temproal understanding. Inside ST-MFGCRN, we devise sentinel attention to calculate the areal similarity matrix by allowing each area to take partial attention if the feature is not useful. We evaluate the proposed model on two real-world transportation datasets, one with our constructed BusDJ dataset and one with benchmark TaxiBJ. Results show that our model outperforms the state-of-the-art baselines up to 7\% on BusDJ and 8\% on TaxiBJ dataset.
Paper Structure (20 sections, 6 equations, 2 figures, 4 tables)

This paper contains 20 sections, 6 equations, 2 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Spatio-temporal prediction methodology
  4. Transit prediction with extra knowledge
  5. Multi-graph spatio-temporal methodology
  6. Problem formulation
  7. Method
  8. Overview
  9. Spatio temporal embedding (STE)
  10. Multi-feature aware GCGRU (MFGCGRU)
  11. Weighted fusion network
  12. Objective function
  13. Evaluation settings
  14. Dataset
  15. Experiment settings
  16. ...and 5 more sections

Figures (2)

  • Figure 1: Similar travel patterns of areas with the same POI and demographics and another data layer of land use (color represents the majority group).
  • Figure 2: (a) Proposed ST-MFGCRN architecture and (b) MFGCGRU cell. TEQ, TEP, TEC represents the temporal embedding (TE) for trend, period, and closeness modules which convert into spatio-temporal embedding (STE).