Overview
SF-LIFE is introduced as a large-scale simulated movement dataset designed for research in transportation, mobility, and machine learning. This dataset comprises 3,024,000,000,000 location records, documenting the complete, noise-free, multi-modality trajectories of 500,000 simulated agents. The agents are observed at a 1Hz frequency navigating the San Francisco Bay Area network over a 70-day period.
Research Context
The dataset aims to accelerate research by providing a robust resource that overcomes limitations associated with real-world tracking data, such as privacy concerns, noise, and completeness issues. SF-LIFE offers an ethically sourced alternative for studies requiring extensive movement data.
Approach
The development of SF-LIFE involved two primary components:
- Generation of needs-driven daily agendas for individual agents through an agent-based simulation of human patterns of life.
- Simulation of detailed kinematic trajectories, moving agents across the OpenStreetMap representation of San Francisco. This simulation utilized data derived from over 40 transit agencies across 9 counties.
The dataset's trajectories are based on real transit infrastructure, incorporating San Francisco General Transit Feed Specification (GTFS) data. Agent movements across multiple modalities are represented, including bus, rail, bike, automobile, and walking.
Findings
The SF-LIFE dataset provides a high-fidelity simulated representation of San Francisco, offering several distinct components:
- Full trajectory data, where each trajectory is annotated with transportation mode labels.
- Reduced-size versions of the trajectory data, available with reduced temporal frequency.
- Agent activity information, which describes the causal activity underlying an agent's visit to a particular place.
- Agent demographic data.
- The underlying OpenStreetMap (OSM) road network and building data for the simulated environment.
The scale and detail presented in SF-LIFE are noted as unprecedented, particularly in its capacity to represent agents' movements across real transit infrastructure and multiple modalities.
Why This Matters
SF-LIFE's design allows it to address limitations inherent in real-world tracking data, specifically concerning privacy, noise, and completeness. By providing a large-scale, detailed, and ethically sourced dataset, it supports research endeavors in transit optimization, human mobility analysis, and urban computing.