MM_DyGNN

This is the python code and data of the article 'Adaptive dynamic graph learning for forecasting urban demand for multimodal travel' submitted to Taylor & Francis in International Journal of Geographical Information Science.This work proposes MM-DyGNN, a novel model for urban travel demand forecasting. It addresses the challenge of capturing complex and dynamic spatial-temporal dependencies from multimodal data sources (bus,metro,taxi). MM-DyGNN adaptively learns graph structures and utilizes a dynamic graph neural network to effectively model the evolving relationships between different transportation systems.