摘要
针对DBSCAN方法存在的参数Eps和MinPts需要事先人为输入及对密度分布层次大的数据集聚类效果较差的局限性,该文对其进行了文献回顾,总结了国内外学者们的研究现状与发展,并比较分析了引用量较高方法的优点和不足,最后得出结论。对于参数确定的问题,现有学者提出了大致两种解决方法:①利用启发式方法;②与其他智能算法相结合。对于具有较大密度差数据集的适用问题,现有学者也提出了大致两种解决方法:①利用曲线斜率将数据集分层;②利用一定的规则将数据集网格化。
Aiming at some limitations that the parameters,Eps and MinPts,need to be input by human previously,and its performance on clustering effect is barely satisfactory for high density distribution levels in the density-based spatial clustering of applications with noise(DBSCAN)method,this paper reviewed literatures on the limitations of DBSCAN and summarized the research status and development of domestic and foreign scholars.In addition,this paper compared and analyzed the advantages and disadvantages of the higher citation methods.Finally we can make some conclusions.For the issue that how to determine the parameters,there has been roughly two solutions proposed by the current scholars:①using heuristic methods;②combining with other intelligent algorithms.For the application of large density difference data sets,the current scholars have put forward roughly two solutions:①using the curve slope to divide the data set;②utilizing certain rules to mesh the dataset.
作者
伏家云
靖常峰
杜明义
FU Jiayun;JING Changfeng;DU Mingyi(School of Geomatics and Urban Information,Beijing University of Civil Engineering and Architecture,Beijing 100044,China;Beijing Key Laboratory of Urban Spatial Information Engineering,Beijing 100038,China)
出处
《测绘科学》
CSCD
北大核心
2018年第12期50-57,共8页
Science of Surveying and Mapping
基金
北京市自然科学基金项目(41771412)
地理国情监测国家测绘地理信息局重点实验室项目(2016NGCM10)
城市空间信息工程北京市重点实验室经费资助项目(2016203)
北京市高精尖中心科研项目(X18058)
关键词
数据挖掘
空间聚类
密度聚类
聚类模式挖掘
data mining
spatial clustering
density clustering
clustering pattern mining
