Short-Term Prediction On Demand Of E-Hailing Trips Based On BP Neural Network

As the continuous penetration of“Internet Plus”in all walks of life,a fresh impetus has been infused for innovating and upgrading traditional industry.The long-existing difficulty in calling a taxi due to information asymmetry between taxi supply and demand,make it a an area for“Internet Plus”to break through.Thus giving birth to a new format of e-hailing trips,which effectively relieves the information asymmetry between taxi supply and demand.In this paper,characteristics of demand of e-hailing trips and are analyzed by data from a e-hailing trips platform,and a short-term prediction modal for demand of e-hailing trips are proposed,which can provide a technical reference for operations to further improving matching effectiveness on supply and demand of e-hailing trips.The main achievements pf this paper are as follow:Firstly,this paper analyses the research situation about taxi supply and demand in both traditional and“Internet Plus”area,and the traffic short-term predictions through a large number of documents.Then through the operation comparison between traditional taxi and e-hailing trips,further clarify the basis of work in following research.Secondly,we conduct the analysis on time and spatial characteristics on demand of e-hailing trips.According to the characteristics of network data,total demand of e-hailing trips is divided into matching number between supply and demand,and demand gaps,and then definitions are giving of matching degree of supply and demand.Different time frames are given after the analysis on the characteristics of demand of e-hailing trips on workdays and non-workdays,and found the difference on matching degree of supply and demand of e-hailing trips within those time frames.On this basis,the spatial characteristics on the demand of e-hailing trips are analyzed,which provides the reference for short-term prediction.Thirdly,to fulfill the practical value,we take the demand gaps as the variable for short-term prediction,and analyze time correlation,and found the significant correlation of demand gaps between current time and 50 minutes before,as well as the same time in days before.Then we construct a short-term prediction model of demand of e-hailing trips based on BP neural network,and determine its structure according to the above correlation analysis results.Then this modal is applied to make predictions on the short-term demand gaps,the validity of the modal is examined as well.Finally,suggestions are proposed to improve the operation of e-hailing trips based on demand and supply analysis,and directions of future research are specified.