Research On Dynamic Pricing And Capacity Estimation Of Ride-hailing Platforms ——Taking Haikou City As An Example

In recent years,taxi-hailing software has been rapidly popularized in residents' lives.On the one hand,the use of taxi-hailing software by residents saves personal time and money,and greatly alleviates the problem of urban traffic congestion.On the other hand,it also improves residents' travel satisfaction index and urban traffic operation efficiency.As the taxi software platform,which is the intermediate contact between passengers and drivers,how to balance the needs and services of passengers and drivers has gradually become a hot issue of research.In the process of accepting taxi software platform services,passengers pay attention to pricing,while taxi software platforms are It is necessary to deploy personnel to serve passengers according to their needs.Therefore,it is necessary and meaningful to consider the situation of passengers using taxi software to get taxis through the Internet as a queuing service system.It is necessary and meaningful to conduct in-depth research on dynamic pricing and capacity planning issues.This paper uses daily travel order data in Haikou City from July 1,2017 to August 31,2017 to conduct an empirical study on passenger travel behavior and dynamic pricing and capacity planning in the queuing service system.Firstly,through descriptive statistics,kepler.gl geographic visualization,Mean Shift clustering algorithm,etc.,the passenger travel behavior is analyzed and researched from two perspectives of time and space;secondly,the theoretical knowledge of queuing theory and reinforcement learning is used to construct a queuing service model,The service demand model,the service cost model and the application of online learning algorithms optimize the dynamic pricing and capacity planning of the queuing service system that passengers use taxi software to take taxis through the Internet.The research results show that the passenger travel demand changes periodically within a week or a day,and the distance between the departure place and the destination of the passenger travel is generally not too far,and the demand is mostly concentrated in one area.The above findings It is conducive to the taxi-hailing software platform to realize reasonable dispatch of vehicles according to the amount of passenger demand,control the occurrence of difficulties for passengers to get a taxi and difficult for drivers to receive orders,and to maximize the satisfaction of the needs of both drivers and passengers and realize the rational use of resources.In addition,through the goodness of fit test and the confirmation of the service desk,it is verified that passengers use taxi software to get taxis through the Internet.This queuing service system can be established as an M/M/C queuing model,and the pricing of passenger taxis is matched to the needs of passengers.Obeying Gaussian distribution,the service capability and service cost fitting of the taxi-hailing software platform obeys the three-item distribution.The set parameters and the observed passenger waiting time and service time are applied to the online learning algorithm,realizing dynamic pricing and capacity planning Optimization.The conclusions drawn from the research work of this paper have certain practical significance for the operation of taxi-hailing software platforms,and are beneficial to the sustainable development of the market for passengers to use taxi-hailing software through the Internet.