Matrix Factorization-Based Trajectory Completion for Addressing Missing Trajectories in Mobile Crowd Sensing

Abstract In this paper, we propose a matrix decomposition-based method for completing missing historical trajectory data in mobile crowd sensing. We model the data as a matrix, exploiting its low-rank properties to efficiently reconstruct missing values. This approach is effective for addressing incomplete trajectory data.We first validate the temporal and spatial characteristics of latitude and longitude data by creating a matrix and performing low-rank analysis. Next, we mathematically model the missing data completion problem using matrix factorization, incorporating spatial and temporal constraints. Finally, we compare our method with traditional matrix completion techniques, demonstrating its superior performance in enhancing dataset accuracy and robustly addressing location data gaps in mobile trajectories.