Fraudulent Taxi Driver detection in Baghdad City based on DBSCAN and A* Algorithm

In Modern cities, fraudulent taxi drivers who take unnecessary detours to overcharge passengers are a major problem in the transportation sector. Therefore, the analysis of taxi driver behavior is an urgent issue to improve the taxi service in any city. Due to this, an efficient system for detecting a fraudulent taxi driver has been suggested; the system firstly, supports the DBSCAN algorithm to construct the jam-distance graph which reflects the actual distance and traffic jam status together of the road network. Secondly, the algorithm has been supported to find the optimal path between two locations based on a jam-distance graph. Finally, a trajectories similarity has been used to measure the dissimilarity between the actual path of the taxi driver and the optimal path to determine the fraudulent taxi driver. The result shows that the proposed system has similar outlier detection accuracy (0.944) based on the TPR rate as compared with STR (0.945), FraudMove (0.921), and iBOAT (0.922). On the other hand, the proposed system (0.2 sec) has less response time for large inquiries as compared with STR(0.5 sec), FraudMove (0.7 sec), and iBOAT (0.9 sec) since the DFT-RN for building its database for finding an optimal path which saves time.