Anomalous Trajectory Detection Using Masked Autoregressive Flow Considering Route Choice Probability

Taxis play a critical role in public traffic systems, and they deliver myriad travelers with convenient service due to temporal-spatial availability. However, anomalous trajectories such as trip fraud often occur greedy drivers. In this study, we propose an trajectory detection method that incorporates Route Choice analysis into Masked Autoregressive Flow, named MAFRC-ATD. The MAFRC-ATD integrates data-driven model-based methods. First, divide the urban network small grids represent subtrajectories sequence of grids. Second, based on subtrajectories, employ model calculate anomaly score each trajectory. Third, according score, can identify distinguish between intentionally unintentionally anomalous. Finally, evaluate our real-world dataset Porto, Portugal. experiment demonstrates effectively discover unintentional detours congestion.