Research On Hot Demanding Spot Analysis And Driving Plan Recommendation For Passenger Searching Based On Taxi GPS Data

Taxis often face the problem of finding passengers in the process of their operation.The unknowingness of taxi demand information increases the blindness and uncertainty of taxi drivers——especially for those inexperienced taxi drivers who often waste a lot of time searching for the next passenger.In order to solve the problem,it is necessary to provide a taxi driver a way to guide their passenger-searching behavior.Therefore,it is necessary to recommend passenger-searching plans to drivers to reduce the blindness while they are searching for passengers.This paper firstly processes and analyzes the historical data from taxi GPS,using SQL Server database to preprocess the taxi data of Shanghai QS taxi company.Then,the pick-up points extraction and OD matching algorithms are respectively set up to obtain the demand information and operation time information of the taxis.Additionally,some operational indicators that reflect the information of taxi demand and unloading state are also analyzed.Secondly,the hotspot of taxi demand is explored based on the processed data.This paper uses the density-based clustering method——OPTICS algorithm,to cluster the taxi demand points.Based on the detailed introduction of the principle of the algorithm,an instance is used to show the optimal parameters' determination and the clustering results.Finally,the core point location of each cluster is calculated as the demand hotspots of the taxis.Finally,this paper comprehensively considers the three indicators of demand hotspot passenger probability,shortest passenger-searching distance and road traffic state index based on vehicle speed.Primarily,the comprehensive recommendation index recommended by the passenger-searching driving scheme is established,and the optimal passenger-searching point is selected to obtain the passenger-searching point sequence.Then,the comprehensive evaluation function of the passenger-searching point sequence is established,and the optimal passenger-searching point sequence is obtained as the recommendation result of the passengersearching driving scheme.In the meantime,the pruning strategy is used to reduce the scale of the algorithm and improve the calculation efficiency.Finally,the example is verified.The results show that the value of the influencing factors corresponding to the optimal search point sequence is in line with the conventional traffic conditions,and the model is effective.