Analysis On Mining Residents’ Activity Patterns And The Change Of Patterns Using Metro Smart Card Data

With the rapid development of social economy,the speed of urbanization construction is gradually accelerated,the urban population is growing rapidly,and the scale of the city is also expanding.A series of urban problems have emerged:disordered urban planning,congested urban traffic,inefficient urban operation and management,imperfect urban service system,etc.Thus the construction of smart city is imminent.As the main body of the city,it is an important scientific issue to study the spatiotemporal activity pattern of urban residents.At present,the mainstream spatiotemporal data analysis method focuses on the spatial characteristics of activity,and regards the activity of residents as occurring at a time point.However,this abstraction and simplification of activity ignores the persistence of activity.In the study which consider the start and end time of activity,most of them focus on the individual activity pattern,and few of them focus on the group activity pattern.For these reasons,this study takes smart card data of Shenzhen Metro as an example,constructs time area of interests to express the start time,end time and duration of residents’ activity,and mines the residents’ activity patterns based on the CLIQUE algorithm with adaptive parameters.Finally,an analysis method based on the center of time area of interests is proposed to analyze the change of residents’ activity patterns.The main research contents and conclusions are as follows:(1)Resident activity pattern mining based on adaptive parameter CLIQUE algorithmIn this paper,the concept of Time Area of Interests is introduced to show the start and end time and duration of residents’ behavior in a two-dimensional plane coordinate system.Due to the large differences in data volume between different dates and the same area,and between the same date and different areas,this paper adopts a CLIQUE algorithm based on adaptive parameters to mine the activity patterns of residents in a long period of time.The clustering results show that the clustering algorithm based on the adaptive parameter setting can effectively mine the significant residents’ activity patterns in different orders of magnitude.There are four types of significant resident activity patterns in this data set,which are work activity pattern,morning entertainment pattern,afternoon entertainment pattern and evening entertainment pattern.Among them,work activity mode and afternoon entertainment pattern are the most common patterns of residents.(2)Analysis on the change of residents’ activity patterns based on the center of Time Area of InterestsIn view of the fact that there are many CLIQUE clustering result graphs when the data period is long,it is difficult to carry out the analysis on the change of residents’ activity patterns or the analysis on the commonness and difference of activity patterns between different regions,this paper presents an analysis method which is based on the center of Time Area of Interests.Through this method,it is possible to effectively analyze the appearances,disappearances,aggregations,and divisions of residents activity patterns in a long period of time.At the same time,the analysis of the changes in residents activity patterns in specific areas is realized,which reveals the differences and common characteristics of residents activity patterns in different types of functional areas,as well as realizes the detection of abnormal patterns.