User Classification And Activity Feature Analysis Based On Weibo Check-in Data

In the era of big data,location-based social media check-in data is increasing and accumulating every day,having formed a large amount of data.These data contain rich spatio-temporal information,which are generated by users on their own initiative.Their work,life and other behaviors will be prominently reflected in them,so they have high authenticity and great mining potential.Exploring the check-in data of social media to analyze the difference of activity characteristics between local and non-local user groups can not only help researchers better understand the urban operation and find out the problems of urban development,but also help decision-makers to allocate resources rationally,make more effective use of urban resources and promote the harmonious development of cities.In order to identify the differences between local and non-local types of users in Weibo,in this paper,the users who checked in Weibo in Shanghai from January 1,2019 to December 31,2020 and their check-in data for the whole year have been obtained.Firstly,the types of local users and non-local users were defined,and the check-in features of Weibo users were extracted from three angles of time,space and quantity to construct feature vectors.Then,the user feature vectors were cross-trained and verified by 10 machine learning algorithms,and the classification accuracy was compared,so as to obtain the optimal classification model for dividing Weibo user groups.Based on the classification results of users in Weibo,the spatio-temporal characteristics of users' check-in activities are analyzed with the check-in spatio-temporal information,the differences of user check-in activities are analyzed with POI types in Weibo,and the location preference of users' check-in activities is analyzed with POI check-in quantity as the index.The following conclusions are obtained:(1)In terms of time dimension,in 2019,local users in Shanghai maintained a low activity intensity from January to June,and increased significantly in the second half of the year,with the highest check-in quantities in December.From January to March,the non-local users' check-in quantities was low.After that,the activities of visiting Shanghai in each month were in a steady state,slightly higher than those in other months in August and October.Affected by the COVID-19 epidemic,in the first half of2020,local users tend to have frequent activities in the city,with the highest check-in quantities in May and stable fluctuations in the second half of the year.Non-local users rarely visited Shanghai in the first half of the year,and began to recover after the May Day holiday,and reached the highest check-in quantities in October.(2)In terms of space dimension,the activity spaces of the two groups of users in Shanghai overlap in many places.Local users have more and wider activity spaces and more check-in quantities in Weibo.The activity hotspots of non-local users are more concentrated,and most check-in hotspots are located in the city center.In 2019,the users' check-in quantities in Weibo increased by 63.01%,and the number of POIs increased by 37.88%,so the space of check-in hotspots is larger.In 2020,the active hotspots of local users are similar,but the hotspots of non-local users are obviously reduced.(3)In terms of activity type preference,the check-in of local users is relatively high in scenic spots and business service types,accounting for 29.79% and 24.35%respectively in 2019,26.35% and 38.81% in 2020,followed by the types of transportation and public facilities;The highest check-in activities of non-local users are scenic spots,accounting for 43.69% and 42.48% in 2019 and 2020 respectively,followed by transportation facilities,accounting for 26.33% and 24.13% in two years,and other types have less check-in.From 2019 to 2020,the activity enthusiasm of both types of users declined,and local users reduced the check-in of scenic spots and preferred the business service type.The preference of non-local users for check-in types has not changed much,and the number of check-in types for business services has also increased significantly.(4)In terms of activity location preference,POI of scenic spots is a hot spot of common interest among users in Weibo.Local users prefer country parks,riverside green spaces and other urban park squares,while non-local users prefer famous tourist attractions in Shanghai.In 2019,local users prefer universities,student dormitories and residential areas.In 2020,local users prefer all kinds of park squares and popular commercial centers.Non-local users' preference for activity location is similar in 2019 and 2020.Besides landmark tourist attractions,there are also important transportation hubs and various accommodation hotels.Taking Shanghai as an example,this paper puts forward a classification method of users' local and non-local types based on Weibo check-in data,and analyzes the spatiotemporal characteristics and activity preferences of the check-in behaviors of the two types of users.The research results are helpful to understand the activity rules of urban residents and tourists,and put forward corresponding optimization strategies for urban construction,so that it can be extended to other super-large cities.