Revealing Collective Human Mobility Spatiotemporal Pattern And Its Driving Mechanism From Taxi Trips Data

Collective human mobility spatiotemporal pattern is a collection of time distribution characteristics and spatial distribution characteristics of travel behavior,which reflects the spatiotemporal dynamic regularity of urban residents' travel behaviors,observing the relationship between residents and land-use in urban environment.Currently,China is in the process of rapid urbanization,therefore the land-use and spatial structure within the city are undergoing profound changes.Aiming at the study of the Spatiotemporal patterns of crowd travel,obtaining the information of spatial structure and social property from the perspective of residents' perception,which provides a new perspective for the study of urban spatial structure and functional structure.Benefit by the development of location-based services(LBS)technology,the travel trajectory data we obtained has significantly improved in terms of quality,scale,and information density,which breaking through the limitations that traditional survey data can only analyze the individual travel behavior,and support the research on the spatiotemporal dynamic characteristics of residents collective travel in urban area.Taxi is an important part of urban public transportation,therefore,exploring the movement patterns of people through taxi trajectory data are important ways to understand the spatial structure and functional differentiation of urban areas,which becoming a research hotspot of GIS and related disciplines.Previous studies have analyzed the spatial heterogeneity of taxi travel demand,but there is still no relevant research on the cyclical changes of travel behavior in the time dimension.The analysis of spatiotemporal differences in travel behavior can reflect the characteristics of residents'travel under more elaborate time and space granularity,and provide an important reference for urban planning,traffic assessment and abnormal monitoring.Based on the analysis of the relationship between taxi trajectory data and road network,this paper discusses the taxi trajectory data set method that takes into account the road network constraints,and constructs the taxi travel time series data set.Starting from the time series characteristics of the taxi travel spatiotemporal pattern,the time series clustering method is combined to identify the regions with similar travel patterns.The random forest is used to construct the association model of travel pattern and POI data,and analyzing the driving mechanism of travel pattern quantitatively.The main contents and conclusions of the research are as follows:(1)Taxi trajectory dataset processing method for travel pattern analysisFirstly,based on the structural characteristics of taxi trajectory data,this paper determines the cleaning and extraction method of taxi OD data;Secondly,the time distribution characteristics of the OD data are analyzed,and the OD frequency time series is determined as the main feature of the regional spatiotemporal pattern;Further,the relationship between the OD point and the road network is analyzed.The research unit is constructed by using the road intersection buffer to collect the taxi trajectory data,and obtain the taxi travel time series data set to lay the data foundation for the spatiotemporal travel pattern extraction.(2)Spatiotemporal Pattern Recognition and Extraction from Taxi Travel Data based on Time-Series ClusteringIn this paper,based on the time series morphological characteristics of the taxi travel pattern in research unit,the time series similarity measure function combining the distance metric and the growth trend similarity is used to calculate the time series similarity between the research units,combining density-based clustering algorithm,a travel time-space pattern recognition method suitable for taxi data is proposed.This method can effectively eliminate the interference of the noise data,identify the travel pattern with similar timing characteristics,and improve the recognition accuracy of the travel pattern.(3)Research on the Driving Mechanism of Taxi Spatiotemporal Travel PatternThe spatial distribution range of the same travel pattern often has the similar characteristics of land-use type and regional function.Therefore,this paper uses POI data to construct the feature set of travel pattern driving mechanism,and establishes the association model of feature set and travel pattern by using random forest.Furthermore,the driving effect of regional function on travel pattern is verified by the classification accuracy of the model.By calculating the feature importance and feature contribution,the research method to quantitative analysis of travel pattern drive mechanism is given.(4)A Case Study of Revealing Taxi Travel Spatiotemporal Pattern and Its Driving Mechanism in NanjingTaking the taxi trajectory data of Nanjing City in September 2010 as the experimental data,the above method was used to reveal the spatiotemporal travel patterns of taxi travel on working days and weekends,and the the spatial distribution characteristics of the travel pattern are analyzed with the electronic map and the land use planning map of Nanjing.Combining the POI data of Nanjing to construct the driving mechanism feature set,which verified that the regional function has a significant driving effect on the travel pattern.Further,calculate the feature importance and feature contribution,and quantitatively analyze the driving mechanism of the POI feature on the travel pattern.