Research On Hotspot Region Mining And Application Based On Taxi Trajectory

With the improvement of people’s living standards,the demand for travel services is also growing.Convenient and quick taxis have become an indispensable means of transportation for public travel.It is convenient for people to travel that a large number of taxis appear,but it also faces problems such as uneven scheduling of public resources,increased energy consumption,traffic congestion,and so on.At this time,it is worth noting that GPS equipment is widely used in taxis.A lot of trajectory data generated by GPS equipment installed in taxis contains abundant information such as time,location and driving status.Exploring taxi trajectory data,and obtaining traffic hotspots to understand travel demand distribution can provide reasonable passenger and boarding information for taxi drivers and passengers,guide drivers and driving novices to avoid crowded road,so as to optimize urban transportation resource allocation,to provide decision support for intelligent transportation at the same time.The current research on trajectory data exploring usually focuses on passenger behavior pattern,driver-seeking preference analysis and traffic flow prediction.In the application of hotspot area exploring,the recommendation of passenger boarding location is not considered in most studies.Driving route planning,real-time navigation and current road conditions are more considered,while less research has been done on the prediction of congestion in unreachable sections.In view of the shortcomings of the current research,in this paper,the author proposes a hotspot region discovery algorithm based on R-DBSCAN,and based on that,carries out the recommendation of passengers and boarding positions,and the author further provides a dynamic road network path planning considering the time factor for taxi drivers to drive passengers.The main contributions of this paper are as follows:(1)Taxi hotspot area exploring In this paper,the taxi trajectory is the research object,and the hot spot area Mining algorithm based on R-DBSCAN is proposed.By exploring the location of getting on and off from taxi trajectory data,using spatial clustering DBSCAN algorithm to cluster according to different characteristic time periods,and using Reverse Nearest Neighbor(RNN)reverse neighbor query algorithm to determine the core hot areas of corresponding time periods,to provide candidate areas for the following location recommendation.(2)Recommendation of taxi passenger and passenger boarding position From the perspective of taxi drivers carrying passengers and passengers taking taxis,the author analyzes the recommendation index for actual passenger loading and boarding in hot area of taxi.On the one hand,the recommended value of the hot spot area is estimated according to the no-load cost and attraction value of the taxi driver when searching for passengers;On the other hand,according to the mobile cost and attractiveness of passengers arriving at the candidate area,the recommendation value of passengers’ taxi taking probability in the candidate area is predicted,and thus the recommendation based on the location of the hot spot area is realized.(3)Taxi Driving Path Planning Because of the complexity and changeability of traffic roads,in this paper,dynamic route planning in which time factors are considered is proposed.With the number of segment state vectors,the congestion coefficient is calculated.Through the time series,the congestion coefficient is predicted,and the heuristic function in the A* algorithm is designed.According to the predicted arrival time,the dynamically changing road congestion coefficient is selected and the path is planned.(4)Implementation of prototype system The prototype of the management and application analysis system based on taxi track data is realized by synthesizing the research results of this paper.Based on the real taxi trajectory,the hotspot areas obtained by the proposed hotspot area exploring algorithm are verified by the prototype system,which are consistent with the actual traffic hotspot distribution.The validity of the system passenger and board recommendation algorithm is proved in comparison with the distance cost of considering only the hot spot attraction as the recommended passenger algorithm.Finally,by compared with the path navigation of Baidu map and Gaode map,the dynamic road network path planning of this system is more inclined to select the road section with good road conditions,which achieves the effect of reasonable avoidance of traffic congestion.