Integrating Knowledge and Data Driven Methods to Generate Synthetic Mobility Data

Urban planning relies on the integration of mobility data from various sources. Analyzing this data can enhance decision-making for transportation services and improve multimodal transportation planning, thereby improving residents’ quality of life. However, traditional data collection methods are often error-prone, costly, and raise privacy concerns. Synthetic data offers a scalable and privacy-preserving alternative for filling gaps in real-world datasets. This paper proposes a hybrid approach that combines knowledge-driven and data-driven methods to generate realistic mobility trajectory data. Mobility data from multiple sources, including demographic statistics and geospatial information, is aggregated and used to simulate traffic patterns with the SUMO traffic simulator. A GAN-based model, specifically CTGAN, is then employed to refine and expand the dataset by generating additional synthetic mobility trajectories. The integration of multiple data sources in this study enhances the availability, diversity, and representativeness of mobility data, supporting more effective urban planning and transportation analysis.