Research On The Method Of Trading Volume's Prediction Of Online Ride-Hailing Based On Deep Learning

The cruising taxi industry is one of the main forces of the whole urban transportation service system.However,there is growing problem of difficulty in calling a taxi caused by the asymmetry of supply and demand information between drivers and passengers.The concept of "Internet +" has provided a new direction for the reform and innovation of traditional industries.Moreover,online ride-hailing service industry has begun from this.Online ride-hailing service effectively eliminates the disadvantages of the asymmetry of supply and demand information between drivers and passengers.On the one hand,it is a great supplement to the transportation capacity of the urban transportation system,which alleviates the travel pressure of urban traffic;on the other hand,it gives people a new understanding and experience of the way to travel.Through the research on the travel data of online ride-hailing service,the purpose is to rationally allocate the transportation resources,so that they can achieve the best match with the travel demand under the premise of satisfying people's daily travel needs and improve the passenger's travel experience.Therefore,it is of great practical value to study and analyze the trading volume of online ride-hailing and its influencing factors.This thesis takes the order of online ride-hailing DIDI service as the main data source,and analyzes the travel characteristics and trading volume's prediction of online ride-hailing based on deep learning and visualization methods in great detail.That provides effective data support for online ride-hailing's resource scheduling,which helps to improve overall operational efficiency,and is of great significance for increasing tranction rate and improving passenger travel experience.First of all,this thesis introduces the travel data of online ride-hailing,and combines the basic methods of related visualization to do visual analysis.Secondly,the daily travel of online ride-hailing is divided into the working day's travel demand and the non-working day's travel demand.The demand matching and shortage degree of online ride-hailing travel is conditional on the travel demand and the unsatisfied quantity.The characteristics of the travel time of online ride-hailing on the working day and the non-working day in different time periods are analyzed.Based on that,the spatial characteristics of online ride-hailing in different travel areas are analyzed.Finally,this thesis Finally,based on the actual demand,the correlation analysis of the factors affecting the trading volume of online ride-hailing is carried out.According to the time characteristics and correlation analysis of online ride-hailing data,it predicts the trading volume by the exponential smoothing model,the RNN neural network model and the LSTM neural network model.And models are verified and evaluated by the actual transaction data.The experimental results show that the LSTM neural network prediction model has the best predictive effect on the trading volume of online ride-hailing,which can reflect the trend in change of the trading volume very well.In addition,the driver resources and passenger travel demand of two important factors affecting the trading volume are analyzed in detail,and the strategy analysis of increasing the trading volume of online ride-hailing is carried ouFurther more,it has certain guiding significance for the operation of online ride-hailing.