Research On Demand Forecast Of Public Bicycle Users From The Perspective Of Resident Green Travel

With the continuous development of economy of China,the accompanying environmental problems have become cannot be ignored.It not only blocks urban and rural development,but also affects the quality of life of residents.In recent years,the operation and development of public bicycles in various cities in China has made the concept of green travel popular,but the problems hidden in it have gradually emerged,and “difficult to rent,difficult to return” has also become one of the urgent problems in current research.By forecasting the user demand of local public bicycle system,the demand of user can be fed back to the system,so as to improve the scheduling efficiency of the public bicycle system,and then the problem of "difficult to rent,difficult to return" can be alleviated.For the public bicycle demand forecast of users,this article cleans the data set of the Citi bike New-York Public Bicycle System(PBS)in April 2019,and divides it into a rent station data set and a return station data set according to the use type of the public bicycle.In order to ensure the scientific validity of the research,the monthly return data and monthly rent data of public bicycles at each station in PBS are distributed and counted.The stations are divided into three frequency bands:high frequency,medium frequency and low frequency.The data of each station in rent station data set and return station data set are loaded into their corresponding frequency bands,and then six frequency band data sets are formed.In order to explore the effect of lending behavior and return behavior of user,this paper divides the research into two main lines: the travel demand prediction of user under the same sliding window and the travel demand prediction of user under different windows.In the study of user travel demand prediction under the same sliding window,this paper explores the effect of user travel demand prediction under different penalty coefficients and slack variables with the support vector regression algorithm and the sliding window method.It is found that the effect of demand forecasting has a bigger affected by the change of the penalty coefficient.In the study of user travel demand forecasting under different sliding windows,support vector regression algorithm is combined with variable sliding windows to analyze the effect of demand forecasting under different sliding windows.When the forecasting effect is similar,the window width and sliding step length show the opposite trend.This study provides practical and effective measures for handling user travel records,and reduces the impact of human factors.The research conclusion not only provides experience for the actual of user travel demand prediction,but also reveals the evolutionary relationship hidden between the demand prediction and the parameters.