Cluster center initialization algorithm for K-modes clustering

作者：

Highlights：

• We address the problem of random initialization of centers of K-modes algorithm.

• We propose an algorithm that performs multiple clustering based on attribute values.

• The proposed algorithm yields fixed initial centers and has loglinear time complexity.

• The algorithm is compared with 3 methods of initialization and shown to perform better.

摘要

•We address the problem of random initialization of centers of K-modes algorithm.•We propose an algorithm that performs multiple clustering based on attribute values.•The proposed algorithm yields fixed initial centers and has loglinear time complexity.•The algorithm is compared with 3 methods of initialization and shown to perform better.

论文关键词：K-modes clustering,Cluster center initialization,Prominent attributes,Significant attributes

论文评审过程：Available online 16 July 2013.

论文官网地址：https://doi.org/10.1016/j.eswa.2013.07.002