A RISK CONTROL FRAMEWORK FOR MORTGAGED CARS BASED ON TRAJECTORY MINING

Feng Lin, Rongrong Jiang, Dongxian Shi, Dong Ren, Mingqi Lv, and Tieming Chen

Keywords

Risk control, mortgaged car, trajectory mining, anomaly detection

Abstract

The car mortgage is a loan, which uses the car as collateral. Thus, the risks of mortgaged cars themselves significantly affect the risks of the car mortgage. Existing work tries to control the risk of mortgaged cars by detecting the abnormal trajectories of the car. However, they would easily lead to false positives because some abnormal trajectories should not be interpreted as risky behaviours. Aiming at this problem, this paper proposes a risk control framework for mortgaged cars based on trajectory mining. Different from the existing work, the proposed framework considers multiple factors to measure the risky level, including the trajectory anomaly, the behaviour anomaly, and the profile data of the borrower. Specifically, the behaviour anomaly is a novel factor that has not been considered by previous works. It extracts the daily behaviour patterns based on frequently visited place mining and detects the behaviour anomaly by matching the current daily behaviour to the behaviour patterns. We conducted experiments on real trajectory data of mortgaged cars, and the results demonstrate the effectiveness of the proposed framework.

Important Links:



Go Back