A clustering method for check-in location data considering location repetition and density peak points

Abstract

The present invention relates to a check-in location data clustering method that takes into account location repetition and density peak points, including the following steps: 1. Constructing the original check-in object and performing preprocessing; 2. Counting the number of original check-in objects at each location to construct 3. Calculate the cut-off distance dc; 4. Calculate the local density ρi and the high-density nearest neighbor distance δi of each FPi; 5. Calculate the thresholds ρ0 and δ0 and filter out the density peak points; 6. Get The core point cluster connected with each peak point density; 7. Search the boundary points of each core point cluster and add them to the core point cluster to form a peak density cluster to complete the entire clustering process. The invention fully considers the problem of repeated check-in locations, effectively avoids the situation that outliers with a high number of repeated locations are selected as peaks and clustered, and the clustering results are more accurate and reliable, and can better reflect the spatial aggregation and activities of urban residents Changes have high practical value.

Description

technical field [0001] The invention relates to a method for clustering check-in location data in consideration of location repetition and density peak points. Background technique [0002] With the popularity of mobile devices with location-based service functions such as smartphones and tablet computers, the location-based social network LBSN continues to grow, providing a good data source for exploring urban business districts and solving problems such as urban transportation and resource allocation. Location check-in is a representative function in LBSN, which means that users use LBS-enabled devices to record their current location, facial expressions, photos and other information and post them on social networks. Because it is very difficult to obtain the accurate location of the user, the existing LBSN generally has a location candidate module, which lists the known locations that the user may be in for the user to choose. Therefore, when different check-in behaviors...

AI Technical Summary

Problems solved by technology

However, most of these methods directly use the spatial distance of point objects as the similarity measure for clustering, and do not consider the repeatability of elements in spatial position.

Directly using the above method to cluster the check-in data is likely to cluster all objects in a certain repeated position into one category, too few points are not conducive to the spatial expression of active hot spots

Moreover, this type of method regards the inside of the density cluster as having a uniform density, and cannot obtain important information such as the density peak, so it is impossible to understand the central tendency of the check-in behavior

[0005] In 2014, Rodriguez et al. proposed a fast search and find density peak clustering algorithm (CFSFDP), but for check-in data with repeated positions, it is easy to select outliers with high position repeatability as density peaks, resulting in inconsistent results. Reliability declines; in addition, since the clustering process has a division method based on the density threshold, the connectivity between the peak and the core point is not considered, and it is difficult to ensure the continuity and integrity of the density cluster

Images