| In 2016,the rapid rise of online car-hailing in various cities caused significant changes in the structure of the taxi market.Since then,taxis based mainly on cruises and online car-hailing based mainly on line orders have jointly occupied the taxi market.Based on this market trend,the newly revised "Taxi Operation Service Management Regulations" clarified the relevant regulations,in order to distinguish the online car-hailing,the full name of the taxi was changed to the cruise taxi,and the transportation management department referred to it as the cruise car.Under the market competition of online car-hailing,the operating efficiency of the cruiser needs to be improved urgently.Therefore,this article will take the cruiser as the research object to study its influence mechanism,demand forecasting and regional dispatch to improve the market competitiveness of the cruiser.First of all,this paper uses python to preprocess and extract OD of taxi GPS data in a certain month in Xi’an in 2017,and use technical means such as data mining and nuclear density hotspot visualization to study the characteristics of taxi travel demand from the time and space dimensions.Analysis shows that there is a significant imbalance in the temporal and spatial distribution of taxi travel demand.Secondly,this paper divides the research area from the perspective of spatial correlation,and finds that the grid division method can better reflect the spatial correlation characteristics of taxi travel demand,so the research area is divided into 487 spatial units.Then select 18 variables that affect travel demand from OSM data,POI data and mobile phone signaling data,and use the GIS platform to fuse the variables with the divided surface files to analyze their spatial heterogeneity.The demand for taxi travel and the Moran index of each variable are calculated.The results show that the univariate Moran index is 0.3 to0.7,and the bivariate Moran index is 0.127 to 0.660,indicating that there is space for taxi travel demand and 18 variables.Correlation,and there is a spatial correlation between the two.Subsequently,the above variables were tested for multicollinearity and a stepwise regression method was used for significance screening.Thirdly,this article uses the GTWR model to analyze the mechanism of the spatiotemporal impact of online car-hailing demand.The fitting equation is used to predict the demand for taxis in different time and space.The results of the model show that the influence of each independent variable on the demand for taxi travel has significant temporal and spatial heterogeneity.This paper selects the ordinary linear regression model(OLS)and the geographically weighted regression model(GWR)as a comparison.It is found that the R2 of the GTWR model is 0.106～0.223 higher than the former two,and the prediction accuracy evaluation indicators RMSE and MAPE are respectively lower than the former two.It is smaller than 83～160 and 8～56,indicating that the GTWR model is more suitable for explaining the mechanism of taxi influence and demand forecasting after considering the spatio-temporal non-stationarity between variables.At last,this paper establishes a taxi regional dispatch model based on the hotspot set.The model is solved by the topological road network and genetic algorithm,and the forecast data of the GTWR model is used for calculation.Calculation example design.The example results show that the model can pre-configure scheduling points according to the demand forecasting,and can meet the demand for taxis in hot spots in the shortest travel time. |
PDF Full Text Request