Time-Space Distribution Forcast Of Public Bicycle Traffic Flow

Nowadays,most cities in China are facing a series of "urban diseases" such as road congestion and air pollution.Bicycles have become increasingly popular as a means of transport because they are healthy and environmentally friendly.The emergence of public bicycles in recent years has served as an extension of urban public transportation,effectively solving the "last mile" problem.However,the operation of urban public bikes system may have the problem of“idle bikes" or“deficient bikes".The former problem will cause the waste of resources,and the latter will cause a bad user experience.Therefore,it is an urgent problem for the government and enterprises to accurately predict the amount of borrowed and reclaimed bikes for each public bicycle rentalBased on the research results at home and abroad,this paper comprehensively analyzes the main factors affecting the amount of borrowing and reclaiming from public bicycle rentals.It analyzes the time series characteristics of the demand of public bicycles in Yancheng.The characteristics include the sequence’s stability,nonlinearity,trend,periodicity,and etc.This paper also provides an overview of the current forecasting methods for the demand of public bicycle rentals.Those methods are analyzed and compared with each other in terms of their advantages and disadvantages,applicable conditions and shortcomings when they are used in work.In order to increase the accuracy of predicting the demand of public bicycles,this paper designs and establishes a new method for predicting the needs of public bicycles,namely the fusion method of ARIMA model and time series weighted regression model.The time series weighted regression model mainly solves the problem of insufficient amount of data,and can successfully fit the periodicity of the predicted data;and the problem of predicting the amount of borrowed vehicles and returning vehicles in the "Yong’anxing" public bicycle rental station in Yancheng City As an empirical study,ARIMA,random forests,time series weighted regression models,and ARIMA and time series weighted regression models are used to predict the number of borrowed vehicles and the amount of returning vehicles.The evaluation index of the measurement error is started,and the prediction result is analyzed and evaluated.The prediction performance of the new prediction method proposed in this paper is predicted on the public bicycle rental pile borrowing amount and the return quantity prediction problem.Finally,based on the combined model,a rule is added to adjust the nonlinear part of the prediction result.The experimental results show that the fusion model of ARIMA and time series weighted regression model has high prediction accuracy,can effectively cope with the shortage of data,and effectively improve the robustness of the prediction model.It is very suitable for public bicycle rental.The method of forecasting the problem of vehicle volume and return volume.