Research On Short-term Forecast Of Online Car-hailing Demand Based On Time Series Analysis

As a necessary part of urban public transport travel system,taxi has enriched the urban passenger transport public transport system with its convenience,flexibility and other characteristics,meeting the diverse travel needs,and has gradually become one of the important ways of travel.The asymmetry of supply and demand information in traditional taxi industry leads to the persistence of taxi difficulty.With the continuous penetration of Intemet+in various industries,network reservation taxi(online car-hailing)emerges as the times require,changing the traditional taxi operation pattern,connecting the supply and demand of taxi with the Internet,alleviating the problem of taxi difficulty and improving the traditional taxi industry.Industry supply-demand relationship.This paper makes use of the open data of drip-drip taxi platform to analyze the influencing factors and timing of the demand of online car-hailing.Based on the time series analysis,a model is built to forecast the demand of online car-hailing,which provides a reference for the operation method of online car-hailing.It is of great significance to improve the operation efficiency of online car-hailing and the matching efficiency of the supply and demand of online car-hailing.This paper mainly makes the following research:Firstly,the correlation analysis of factors influencing the demand of online car-hailing appointment is made.Combined with the demand data of online car-hailing,the influence of POI number,weather condition,temperature,PM2.5 and road congestion on the demand of online car-hailing is analyzed,which provides a basis for short-term forecast of demand of online car-hailing.Secondly,the research on the trend of the demand change of online car-hailing.The time series analysis of the demand data of online car-hailing shows that the demand of online car-hailing changes periodically in seven days.The trend of the demaind change of online car-hailing on working days and non-working days is analyzed,and the changing law of the demand for online car-hailing is found out.The different trend characteristics of the demand of online car-hailing on working days and non-working days are obtained as models.Type feature extraction provides a basis.Thirdly,short-term forecasting of online car-hailing demand.Based on the time series of online car demand data and the changing trend of online car-hailing demand data,short-term forecasting models of online car-hailing demand based on ARIMA,single-feature LSTM and multi-feature LSTM are established respectively.The optimal parameters and models of the model are determined by the actual demand data of online car-hailing demand based on drip-drip taxi platform.Structures and comparative experiments are carried out.The experimental results show that the multi-feature LSTM prediction model is the best in the short-term forecasting of the demand for online car-hailing.