Urban business-circle cluster partition method based on self-adaptive DBSCAN density clustering

Abstract

The invention provides an urban business-circle cluster partition method based on self-adaptive DBSCAN density clustering. The method includes the following steps that 1, for all shops of one type ofall cities, the nearest distance is subtracted form 1, the global DBSCAN clustering radius epsilon G is obtained, and the quartile Q of the number distribution of this type of shops of all the citiesis calculated; 2, longitude and latitude data of all this type of shops of the city Ci is obtained; 3, whether the number of this type of shops of the city is larger than the Q or not is judged. If the number of this type of shops of the city is larger than the Q, the city shop clustering radius epsilon is independently calculated, MinPts=1, and DBSCAN density clustering is conducted; otherwise, the global clustering radius epsilon G is used, MinPts=1, and DBSCAN density clustering is conducted. For cities different in shop number and scale, different business-circle cluster partition strategies are performed, the robustness of the geographical location cluster partition result is improved, the business-circle layout characteristics of one type of shops in different cities are effectivelyreflected, and a subsequent recommendation system thus can easily explore the user behavior about the change of interest points in geographical positions.

Description

technical field [0001] The invention relates to the fields of data mining and computer technology, in particular to a cluster division method of urban commercial circles based on adaptive DBSCAN density clustering. Background technique [0002] The design of the recommendation system is more and more focused on finding changes in users' points of interest (POI, Point Of Interest). For example, music applications will pay attention to users' preferences for different types of music; news websites will pay attention to their hobbies for different types of news; and The e-commerce platform will collect which products customers are concerned about, and so on. The recommendation system expects to find out the changes of users' points of interest as accurately as possible, so that the recommendation results can meet the needs of users as much as possible. [0003] Nowadays, Location Based Service (LBS, Location Based Service) has been widely used in people's daily work and life. ...

AI Technical Summary

Problems solved by technology

[0005] In order to overcome the disadvantages of poor robustness of the cluster clustering division effect of geographic location data in the prior art, the present invention provides an adaptive DBSCAN density clustering algorithm with good robustness and good cluster division effect. The cluster division method of urban commercial circles

Images