From GPS to Maps: Path Reconstruction Techniques for Understanding People Mobility with Low-Sampling-Rate Data

Urban mobility is a critical aspect of city planning and transportation management, with the Global Positioning System (GPS) playing an essential role in analyzing and understanding patterns of human movement. Particularly using smartphones, GPS provides highly detailed digital trace data, enabling in-depth analysis of human movement behavior. However, there are significant challenges when working with GPS datasets that have a low frequency, which often occurs due to storage capacity limitations or data transmission costs. This study explores methodologies for reconstructing movement paths from low-frequency GPS datasets. The process includes several key steps: data pre-processing involving cleaning and normalizing data; aligning GPS data with existing road networks using the nearest-neighbor KD-Tree algorithm; and accurate path reconstruction. In addition, the study develops a method for time interpolation estimation to determine movement times between the aligned coordinate points. The findings of this research indicate that with the proposed approach, movement paths can be reconstructed effectively and accurately, even from data with low sampling frequencies. This approach promises a wide potential application, ranging from transportation infrastructure planning to the development of more sophisticated navigation systems. With ongoing research, the integration of additional sensors and machine learning techniques is anticipated to further enhance the accuracy and utility of this methodology.