基于变量加权的软子空间聚类方法数据集

面向成员属性复杂、异质等问题，从相异度和几何空间等角度出发，挖掘成员的空间分布及成员间的关联度。提出了一种基于变量对聚类重要性的变加权k-均值聚类算法(VwKM)。在VwKM中，所提出的可变权重方案可以识别一个聚类的重要变量，也可以为相关专家提供知识和经验。然后，提出了一种基于反向最近邻的密度峰值方法(RP)来解决VwKM的初始化敏感性问题。

To address the problems of complex and heterogeneous member attributes, this study investigates the spatial distribution of members and the correlation among members from the perspectives of dissimilarity and geometric space. A variable-weighted k-means clustering algorithm (VwKM) based on the importance of variables to clustering is proposed. In the VwKM algorithm, the proposed variable weight scheme can identify the key variables for a specific cluster, and also provide relevant knowledge and experience for domain experts. Subsequently, a density peak method based on reverse nearest neighbors (RP) is proposed to solve the initialization sensitivity problem of VwKM.

搜集汇总

背景与挑战

背景概述

该数据集针对复杂、异质的成员属性问题，从相异度和几何空间角度探索空间分布与关联度，提出了变加权k-均值聚类算法（VwKM）以识别聚类中的重要变量，并引入基于反向最近邻的密度峰值方法（RP）来缓解初始化敏感性。它属于人工智能领域，包含111.33MB数据，由武汉大学于2023年发布。