The Visualizing Analysis Of User Behavior Based On "DiDi" Order Track Data

The trajectory data generated by people's travel activities contain their hidden travel behavior characteristics.Through the study of travel trajectory,people can analyze their travel habits and needs,thus helping enterprises to develop a series of products and service that meet people's travel activities.Traditional resident travel behavior data is generally obtained through manual interviews,questionnaires or less precise GPS positioning techniques.The acquisition cost is too high and the data accuracy is low.However,as a leader in the network car industry,”Di Dichuxing” has not only a wide range of platform services,but also a high positioning accuracy of GPS technology,making the starting point of the order closer to the user's real position.In addition,the private car ordering service provided by “Di Di” further increases user coverage.This makes the “Di Di” user more accurately represent the actual travel behavior of a city resident as a whole,so the “Di Di” user behavior can represent the urban population,and research on it can provide more accurate decision-making reference basis for relevant city decision makers.Based on the “DiDi” order trajectory data with high-precision positioning technology and closer to the user's destination,this paper clusters and visualizes the travel rules of “Di Di” users,aiming at “Di Di” users,“Di Di” Drivers,location providers,cities and land planners provide some decision support.Firstly,this paper cleans,processes and extracts the original “Di Di” order trajectory data,and classifies and stores the data for different needs,and climbs the point of interest data of Gaode map as the data foundation of cluster visualization;Secondly,the descriptive statistical analysis is carried out on the “Di Di” order trajectory data in different time periods(week travel volume,sunrise traffic volume,time travel volume,travel time length).Then,the K-Means clustering of the travel time and driving time of the “Di Di” user is carried out,and the user's hot spot travel time and hot travel time length are explored as the basis of the hot spot research,and through the adaptive parameter density clustering algorithm(DBSCAN),the multi-indicator visual analysis of the travel hotspot area of the “Di Di” user is carried out.Thirdly,the population classification model is proposed based on the POI data.The research on the traffic volume,hotspot travel time,average travel time,and travel hotspots of college students,residents and office workers are studied.According to the results of K-Means clustering and DBSCAN clustering,the differences in travel behavior of these three groups were explored.Finally,combined with residential orders and office workers orders,this paper extracted the commuter orders of “Di Di” users,explored the commuting time and occupational situation of office workers in Chengdu,and visualized the cluster analysis results in Arc GIS software,which makes it more intuitive and accurate to analyze the temporal and spatial interaction behavior characteristics of office workers in “Di Di” users.This study has certain theoretical and practical value for revealing the space-time interaction mode of “Di Di” users and optimizing the relevant decision-making in Chengdu.On the one hand,it provides the time and place for the “Di Di” driver to have quality orders,making the driver's decision more efficient;On the other hand,it analyzes the balance of employment in Chengdu and helps urban planners optimize land use decisions.This paper also provides a certain decision-making basis for the marketing activities of location service providers such as scenic spots and entertainment venues through the analysis of the travel hotspots of “Di Di” users.