An Research And Application In Rental Trends’ Forecasting Of Public Bicycle Sites

The traffic and environmental problems caused by the development of the city,is an important problem in many big city’s development.The appearance of PBS can effectively alleviate these problems in traffic and environment.However in the process of using PBS,there is a serious problem that can restrict the development of PBS due to the problem “too hard to borrow,too hard to return” caused by the tidal property in residents.Therefore,it is imperative to study a model to predict precisely the number of available public bicycles.At present,there is less research on the public bicycles’ prediction model,but it is already mature on the motor vehicle flow prediction.So the current research on public bicycle traffic flow prediction mostly comes from the field of motor vehicle traffic flow.Due to the difference between public bicycle traffic flow and motor vehicle traffic flow,the current research can’t accurately predict the public bicycle traffic flow,especially in the study of the long and short term prediction of the number of available public bicycles.This paper have proposed a long-term rental prediction model for public bicycles which combine three factors of “weather,temperature,holiday” by the analysis in the use of public bicycles and rental trends and a short-term rental prediction model for public bicycles based PSO-SVM.The long-term prediction can provide the theoretical basis for scheduling and the short-term prediction can make a precise prediction for users.The main innovations of this paper are as follows:(1)Firstly this paper research the rental trends of public bicycles sites through the analysis in rental behavior,travel time and spatial behavior.Including,(a)this paper have found the difference of residents’ travel in weekend and week day through the analysis of rental behavior.(b)this paper analysis the situation of tidal property in each kind of public bicycle sites through the analysis of travel time.(c)this paper have found the flow situation in residents through the analysis of the spatial behavior.(2)This paper have provided long-term prediction model in public bicycles combining historical data,public bicycle rental records,the data of weather,holidays and temperature.(3)This paper have also provided short-term prediction model by particle swarm optimization SVM.