Research On Taxi Demand Forecasting Model

Due to the rapid development of social economy,taxis have become the preferred way for people to travel.With the development of network technology and the popularization of intelligent terminals in recent years,taxis under the network taxi platform(hereinafter referred to as taxis)by virtue of its fast,convenient and high-quality service gradually become the first choice for people to travel.However,in reality,it is difficult for taxi drivers to know the demand for taxis in different regions of the city at different times,which may lead to the phenomenon of no-load or short supply in taxis,which wastes the social resources.In order to solve this problem,this paper researches the forecasting model of taxi demand and forecasts the demand of taxis at different times in the future.Firstly,this paper studies the common methods of taxi demand forecasting and the related short-time traffic flow forecasting model,and analyzes the characteristics of the common models.By using the data visualization method to verify the weather conditions,PM2.5,temperature and traffic congestion and other factors have a certain impact on the short-time demand of taxi,and then draw on short-time traffic flow forecasting problem to construct the variables method to extract and design the sample,which lays the foundation for the establishment of the model.Then this paper constructs multivariate linear regression model,random forest regression model,gradient progressive regression tree model,and proposes a linear variable weight combination forecasting model on these three models to predict the demand of taxis.Where the weight of the combined forecasting model is determined based on the reciprocal of the mean-square predictive error of the single model,and the weight of the model is adjusted in real time.Finally,taking the data of taxi demand in a first-line city in China as an example,this paper verifies the validity of the combined forecasting model constructed in this paper,and the experiment shows that the overall forecasting accuracy of the combined forecasting model of linear variable weight is higher than that of the single prediction model.In order to ensure the universality and convenience of the model,this paper uses Rserve to realize the taxi demand forecasting system which interacts with the JAVA language and R language,taking into account the response time of the system,the system adopts C/S architecture.