WorldMove, a global open data for human mobility.

Area of Science:

• Computational Social Science
• Geospatial Data Science
• Urban Informatics

Background:

• High-quality human mobility data is essential for urban planning, transportation, and public health.
• Data collection faces challenges due to privacy concerns and scarcity, especially in developing regions.

Purpose of the Study:

• Introduce WorldMove, a large-scale synthetic human mobility dataset.
• Provide a privacy-preserving and scalable solution for global mobility research.
• Enable custom synthetic data generation for any city.

Main Methods:

• Leveraged multi-source data: gridded population, POI maps, and origin-destination flows.
• Employed a diffusion-based generative model to simulate realistic mobility trajectories.
• Developed an open-source pipeline for data generation and model training.

Main Results:

• Generated synthetic mobility data for over 1,600 cities across 179 countries.
• Validated data alignment with real-world individual behavior and city-scale flows.
• Released the trained model and open-source pipeline.

Conclusions:

• WorldMove addresses critical data gaps in human mobility research.
• The dataset and tools promote scalable, privacy-preserving, and inclusive mobility studies.
• Facilitates universal access to human mobility insights, particularly for data-scarce regions.