一种带有虚拟变量的密度聚类算法

A Density Clustering Algorithm with Dummy Variables

具有空间和虚拟变量属性的数据集在现实世界中普遍存在,如卫星数据、气象数据等,对这类数据进行聚类可为近一步科学研究提供启迪。在ST-DBSCAN算法基础上,针对空间虚拟变量属性数据集的聚类问题,提出带有虚拟变量的DST-DBSCAN密度聚类算法。使用ST-DBSCAN无法完成虚拟变量的数据集进行聚类。DST-DBSCAN算法在不改变ST-DBSCAN算法对空间、非空间、时间属性聚类功能和算法时间复杂度的基础上,增加了虚拟变量属性的聚类功能。最后本文以台风为例,使用DST-DBSCAN算法寻找到生成登陆台风可能性较高的海域。

A space and dummy values of the objects are widespread in the real world such as satellite data,meteorological data.Clustering such objects can push scientific research a step forward.In this paper,the new density clustering algorithm,DST-DBSCAN,is based on ST-DBSCAN algorithm,in order to solve space and dummy variables for the attribute data sets clustering problem.It does not change the ST-DBSCAN algorithm space,non-spatial and temporal clustering capabilities and the time complexity,increased dummy variable clustering capability.Finally,this paper uses DST-DBSCAN algorithm to find the possibility of sea areas generating landing typhoon.