Research On Suggestion Of Selecting Online Orders And Recommendation Of Passengers-seeking Area For Taxi Based On Profit Indicator

The combination of "Internet Plus" and transportation has made great changes in the taxi industry.The new way of traveling by network-based car has brought significant impact on traditional taxis,but also a new opportunity to taxi industry for development,as the efficiency is too low due to information asymmetry and random transaction.Whether it is the network-based car or traditional taxi,how to improve the profits has always been the most concern of enterprises and drivers.In order to improve the profits,this paper builds a profit indicator,then makes suggestion for selecting taxi online orders and recommendation of passengers-seeking area based on this profit indicator.Firstly,by analyzing domestic taxi data in current studys and available datasets of domestic taxi,and by comparing taxi travel conditions between domestic first-tier cities and New York City,this paper illustrates the insufficiency of domestic taxi data,and shows the rationality and applicability of using taxi data of New York City in this research.With this data consisting of 13862 taxis in New York City,this paper cleans up abnormal and unreasonable records,and demarcates a specific research area by ray casting algorithm.Secondly,this paper analyzes the characteristics of spatial and temporal distribution based on taxi trips data in the research area.The analysis of temporal distribution is carried out from four angles:travel demand,travel time,travel distance and travel fare.And the difference between weekdays and weekends,as well as the change of different periods in one day are obtained.The analysis of spatial distribution is carried out by travel demand,travel destination and travel direction.The spatial distribution of travel demand and travel destination are studied,and the temporal and spatial characteristics of travel direction are found.Then,considering the influence of supply and demand in area,passenger destinations and other factors to the efficiency and profit of taxi operation,this paper constructs a profit indicator based on Markov Decision Process(MDP)?and makes a detailed study on the calculation of the indicator parameters.According to the characteristics of spatial and temporal distribution,this paper gives a time division method,and also a area division method which is based on K-means++algorithm and could highlight the destination characteristics of the areas,and divides the research area into small areas.Finally,adapting to the development demand of the taxi industry,the application of profit indicator is carried out from two aspects.On the one hand,when the driver of the network-based car receives more than one online oreder,the estimated incomes in a certain time can be calculated by the profit indicator as decision help,and a suggestion of selecting these orders will be offered.On the other hand,this paper established a recommendation model of high profitable passengers-seeking area based on the profit indicator for the cruise car.By simulating the recommendation and operation process,and then comparing the average income and the cruise time rate in the study period of simulation and reality,the validity of the model has been proved.This paper contains 39 figures,16 tables and 60 references.