Floating Population Identification And Its Population Travel Characteristics Analysis Based On Social Media Data

With the advent of the Web 2.0 era and the popularization of social media,population monitoring based on social media data can be a beneficial supplement to the traditional population survey.According to spatio-temporal information of a large number of users' activities,floating population can be identified,and the floating population is further classified.The study of travel characteristics of floating population and their relationship with urban traffic can make urban traffic more targeted for the service of floating population.For the identification of the floating population from social media users,this research firstly defines the floating population and four types of localness: long-term residents,temporary residents,visitors and tourists,and the last three types are considered as floating population.Then,from time,space and social three perspectives,features of social media users are extracted,and conditions and sequence are designed for localness evaluation.Using the Twitter dataset in Greater London,the localness evaluation approach is implemented,and 738 users were labelled as one of the four types of localness.Later,the spatial and temporal information of these users was used to extract the travel characteristics of the four types of localness population.The correlation coefficient,spatial clustering and other methods are used to analyze the travel characteristics of floating population in London from two perspectives of time and space and the relationship between the trips and urban traffic in London,and draws the following conclusions:(1)The floating population in Greater London may have fewer changes in February and May 2018.If only tourists and visitors who stayed less than a week are taken into consideration,July,August and December are the peak periods for these kinds of floating population to enter and leave the Greater London every year,and the floating population will create a great burden on the urban transportation system in these periods.(2)Most of the social media user activities are close to public transportation site locations.Distance between the bus stops and activities have no obvious trend,but the distance between rail transit stations and activities present a descending trend with the decrease of localness,namely,the shorter the population of localness stay,the closer the activity location to the rail transit stations.(3)There is a significant positive correlation between the proportion of tourists' stay time and the traffic volume of subway and bus,with the correlation coefficients reaching 0.69 and 0.52 respectively,while neither the linear correlation between the traffic volume of subway and bus nor the relationship between various floating population and private traffic volume is significant.(4)The density-based clustering algorithm was used to obtain the travel activity clustering centers of all kinds of population.The analysis concluded that the clustering centers of longterm residents,temporary residents,visitors and tourists decreased successively,and the distribution of clustering centers became more and more intensive.The clustering centers of all kinds of population are located in the densely distributed areas of urban public transport stations or along the rail transit lines.