Research On The Public Bicycle Travel Characteristics And Prediction

With the rapid development of urban mobility, the problem of urban transport has been becoming worse. Road congestion and air pollution problems have become common in major cities, which are known as "urban disease". How to develop an efficient, low-carbon transportation is that all of us must confront. The Public Bicycle System (PBS) is considered as a zero-emission and healthy travel pattern and it becomes more and more popular in the world. As a short-distance travel pattern, the characteristics of PBS are energy saving, convenient and flexible, practical and economic. Besides, it can effectively extend the service area of the public transport. However, many problems emerged during the boom in development of PBS, such as difficulties in rental and return bicycle, the imbalances between income and expense, bicycle rebalance and so on, which seriously hinders the development of public bicycle system.In this paper, the main data derive from Citi Bike New York City public bicycle card data. Firstly, the data is used to research PBS potential feature of the trip based on data mining method, from which to find the influence mechanism of main factors. Secondly, the influence factors will be integrated into the prediction model to acquire the optimal prediction model. The significance of this paper is to support vehicle scheduling and user riding, which improve the level of service in PBS. The aspects of researches are as followed:First of all, this paper introduces the basic situation of city which PBS is located in. Besides, it analyzes physical facilities, operations management as well as the problems in detail.Secondly, this study analyzes PBS’s travel features from time-space and users two aspects, from which to mine the influence mechanism of main factors.Thirdly, this article takes the similarity of characteristic curve as the standard to cluster system stations, which is prepared for predicting the number of station Bicycles.Finally, this paper integrates two main factors (weather and the type of station) into the ARIMA mode. The travel in different types of station is predicted respectively, which improves the prediction accuracy.