Resident Travel Activity Pattern Mining Based On Multi-source Spatio-temporal Trajectory Data Clustering

Public transportation is an important part of urban traffic and also a vital vehicle of urban residents to daily travel.Therefore,investigating the travel of urban residents is of great significance for analyzing urban public transit.The study of traditional residents travel relies on travel survey data,but this approach is time-consuming and labor intensive,and the amount of sample obtained is small.With the development of information technology,the intelligent transportation system is widely used in the public transportation system.The vehicle-mounted intelligent sensing device can acquire a large number of time-space trajectory data based on micro-individuals,which not only records the time information of passengers taking public transportation but also records the spatio-temporal details of vehicle trajectory,providing a new way to study the travel mode of residents.This paper based on Shenzhen residents' credit card data,including IC card data,bus GPS trajectory data and basic GIS data.These datasets contain rich information on urban residents' travel activities,reflecting the variety of time and space of urban micro-individual travel activities.Based on these data,the boarding and alighting stops of the passenger can be deduced,and the passenger travel chain can be constructed for the analysis and excavation of residents' travel activities,and various modes and rules of residents' travel activities can be found.The common method of mining travel patterns of residents is to cluster the trajectory data.This method can effectively extract the similarity features in spatio-temporal trajectory data and find meaningful travel rules.The traditional clustering method is not suitable for spatiotemporal sequence data with information of position and time.Therefore,the traditional clustering method needs to be expanded in space and time to meet the research needs.Firstly,based on multi-source data and related rule constraints,we derive the location of the passengers' trips from the original dataset,constructs travel trajectory chain and analyzes the temporal and spatial distribution of residents' travels.Then,using the proposed trajectory similarity measure model to map the residents' travel trajectories Based on the trajectory similarity,the pattern of residents' travel activities is mined by of hierarchical clustering algorithm.Finally,the causes for the formation are analyzed according to the classification results and social background of the clustered travel patterns.This study not only helps to reveal the patterns of resident travel activities hidden behind large-scale trajectory data,but also provides a reference for understanding the interaction between people and cities.