Multi-Modal Urban Mobility Forecasting: A Graph Neural Network-Based Approach with Spatiotemporal Hypergraph Attention

Adabala, Surendra (2025) Multi-Modal Urban Mobility Forecasting: A Graph Neural Network-Based Approach with Spatiotemporal Hypergraph Attention. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (2MB) | Preview
Preview	PDF (Configuration Manual) Download (562kB) | Preview

Abstract

Accurate forecasting of urban mobility patterns is essential for improving traffic management, resource allocation, and operational planning in smart cities. However, traditional models such as ARIMA and standard LSTM architectures of- ten fall short in capturing the complex spatiotemporal dependencies inherent in multi zonal, multi modal transport systems. This study proposes a unified fore- casting framework The Dynamic Spatiotemporal Hypergraph Convolutional Network (DSTHGCN) integrates three established techniques: hypergraph-based spatial modeling, hierarchical attention mechanisms, and GRU-based temporal encoding. Rather than introducing new algorithms, it provides a novel architectural synthesis that addresses key limitations of prior methods, including the inability to model many-to-many zone interactions and long-term temporal patterns.

The model was evaluated on real-world inflow and outflow data from New York City taxi and bike-sharing systems across 69 mobility zones. Comparative experiments show that DSTHGCN consistently outperforms baseline models (ARIMA, LSTM, GCN), achieving Mean Squared Error (MSE) of 0.0041, Root Mean Squared Error (RMSE) of 0.0639, and Mean Absolute Error (MAE) of 0.0438, demonstrating improved accuracy and stability. These results highlight the effectiveness of hypergraph-based and attention-enhanced spatiotemporal modeling for urban mobility prediction. Despite limitations, such as dependence on precomputed hypergraphs, DSTHGCN represents a scalable and interpretable framework for real-time, multimodal demand forecasting in intelligent transportation systems.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Razzaq, Abdul UNSPECIFIED
Subjects:	Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence H Social Sciences > HT Communities. Classes. Races > Urban Sociology > City Planning H Social Sciences > HE Transportation and Communications > Urban Transportation
Divisions:	School of Computing > Master of Science in Artificial Intelligence
Depositing User:	Ciara O'Brien
Date Deposited:	28 May 2026 11:39
Last Modified:	28 May 2026 11:39
URI:	https://norma.ncirl.ie/id/eprint/9310

Actions (login required)

View Item