SynMob

A High-Fidelity Synthetic GPS Trajectory Dataset for Urban Mobility Analysis

Introduction

We present SynMob, a high-fidelity synthetic GPS trajectory dataset for urban mobility analysis. SynMob is generated by a novel diffusion probabilistic model, DiffTraj, which can generate trajectories with high fidelity and diversity. We help this Synthetic GPS Trajectory Dataset can facilitate the community for urban mobility analysis.

Overview Image

Example Trajectories

Trajectory Image 1
Different origin-destination
Trajectory Image 2
Same origin-destination
Trajectory Image 3
Different origin-destination
Trajectory Image 4
Same origin-destination

Dataset Summary

Summary of comparison with existing trajectory datasets
Dataset GPS trajectory Avalibility Data quality Privacy # Trajectory
GeoLife ✓ ✓ ✗ ✗ 17,621
T-drive ✓ ✓ ✗ ✗ 10,357
Porto ✓ ✓ ✗ ✗ 1.7 million
Foursquare ✗ ✓ -- ✗ 104,478
NYC ✗ ✓ -- ✗ 1.1 billion
Taxi-Shanghai ✓ ✗ ✗ ✗ 1.2 million
GAIA ✓ ✗ ✓ ✗ 3.1 million
Ours (Synthetic) ✓ ✓ ✓ ✓ unrestricted (customize)

Citation