Identifying Urban Functional Regions by LDA Topic Model with POI Data

Abstract

AbstractIdentifying Urban Functional Regions (UFR) can achieve the rational aggregation of social resource space, realize urban economic and social functions, promote the deployment of urban infrastructure, radiate and drive the development of surrounding regions, so the identification of urban functional regions can promote the efficient development of cities. However, the traditional functional region identification method is mainly based on remote sensing mapping, which relies more on the natural geographical characteristics of the region to describe and identify the region, while the urban functional region is closely related to human activities, and the traditional functional region identification results are not ideal. Social data includes a series of data that reflect people’s activities and behaviors, such as trajectory data, social media data, and travel data, thus the analysis of social data can more effectively solve the difficulties of traditional mapping and identification. POI (Point of Interest) data, as a typical type of social data, can be used to identify urban functional regions. We apply the LDA topic model to the POI data, and propose a new urban functional region identification method, which makes full use of the POI data to reflect the activity categories of urban populations to characterize the features of regional functions and achieve a high degree of identification of urban functional regions. Through experimental verification on real data, the experimental results show that the proposed method can more accurately identify urban functions, which proves the method reliable.KeywordsUrban functional regions identificationPOILDA topic modelClustering algorithm