Gradient Boosted Regression Tree For Dynamic Demand Prediction Of E-hail Taxis

Recently,E-hail Taxis is wildly used,and the taxi-hailing app has been accepted by the public and the government.As a commonly used transportation,its convenience is also getting higher and higher requirements.The demand forecasting of taxis number is also been seen as an important technology breakthrough by all the taxi-hailing app company.The research on the demand forecast of taxis number is gradually focused by the scholars.At present,the research is mainly concentrated on forecasting the taxi number of a city needed or finding the hot spots of passenger in that city,which means there is large space for the researchers to make up for the lack of the research about forecasting the dynamic demand for E-hail Taxis.Based on given data,predicting the real-time vehicle demand in different regions of a city is the dynamic demand forecasting problem described in this paper.In this paper,the real-time demand of the rental vehicle problem is analyzed.With analyzed a large number of relevant data,approaching a novel forecasting method based on Gradient Boosted Regression Tree algorithm,to solve dynamic demand prediction of E-hail Taxis.this is the first time to use machine learning algorithm to solve demand prediction of E-hail Taxis problem.Firstly,based on the existing historical operating data of the taxi-hailing app and external environment condition,a model is established to forecasting the number of real-time vehicles needed in the urban area at the next time segment.Secondly,according to the actual operation data from a certain intelligent travel platform,the real-time forecasting model is verified.The simulation shows that the models established has great prediction results.