签到位置数据的密度峰值快速搜索与聚类方法

发布时间：2018-03-14 21:26

本文选题：签到位置数据　切入点：活动热区　出处：《测绘学报》2017年04期 　论文类型：期刊论文

【摘要】：位置签到数据蕴含了城市居民活动变化。由于客户端位置候选问题,不同的签到行为以同一候选位置签到时会产生位置重复现象。针对现有密度聚类方法在签到数据聚类上存在的问题,以快速搜索和查找密度峰值聚类算法(CFSFDP)为基础,提出了签到位置数据的密度峰值快速搜索与聚类方法。首先,引入位置重复频率来表达签到位置重复,然后,对原始签到位置数据点统计位置重复频率并重新设计数据结构,以新的空间点要素为研究对象寻找密度峰值点;最后,构建了峰值点密度簇聚类算法,在点要素集聚类过程中考虑密度连通性来保证峰值密度簇的连续与完整。试验表明,所提出的聚类方法有效避免了重复度较高的离群位置对象选为峰值并聚类的情况,并具有良好的空间适应性。所提取的密度峰值点不仅可以用来表示热区的中心,还能够反映热区的集中趋势,进而可以帮助探索热区的动态变化情况。
[Abstract]:Your location data contains changes in city residents. Because the client position candidate problem, sign different behaviors in the same position in the candidate will have to repeat position phenomenon. In view of the existing density clustering method in attendance data clustering on the issue, to quickly search and find the peak density clustering algorithm (CFSFDP) as the foundation, proposed the peak density sign position data fast search and clustering method. Firstly, then introduce the position of repetition to express the original position, repeat sign, sign position data statistic position of repetition frequency and re design of data structure, find the density peak with elements of the new space as the research object; finally, construct the peak point density cluster clustering algorithm in clustering process considering the point density connectivity to ensure the continuity and integrity of the peak density cluster. Experiments show that the proposed poly Such method can effectively avoid the outlier location object high degree of duplication for peak and clustering, and has good adaptability. The extracted spatial density peak not only can be used to express the zone center, the central tendency can reflect the dynamic changes of hotspots, and can help to explore the hot spot.

【作者单位】：福州大学福建省空间信息工程研究中心空间数据挖掘与信息共享教育部重点实验室;
【基金】：国家自然科学基金(41471333)~~
【分类号】：P208;TP311.13

【相似文献】