An improved method for k-means clustering based on internal validity indexes and inter-cluster variance
by Guangli Zhu; Xiaoqing Li; Shunxiang Zhang; Xin Xu; Biao Zhang
International Journal of Computational Science and Engineering (IJCSE), Vol. 25, No. 3, 2022

Abstract: It is difficult to get the best cluster number of k-means clustering algorithm by using the traditional internal validity indexes. Therefore, a good clustering result cannot be obtained usually. To solve this problem, this paper proposes an improved method for k-means clustering based on internal validity indexes and inter-cluster variance. Firstly, some integers distributed in the interval [2, √n] are selected as initial cluster numbers. Further, each initial cluster number is selected to carry out k-means clustering and obtain a clustering result. Secondly, two initial cluster numbers k_D and k_C are extracted respectively under two optimal validity index values (i.e., DB and CH). The k_D and k_C are extracted by comparing and analysing the validity index values. Finally, the validity index ICS-VAR is proposed to select the best cluster number k_B, while k_D is not equal to k_C. Experimental results show that the improved method can obtain a better cluster number under a certain condition.

Online publication date: Mon, 30-May-2022

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Computational Science and Engineering (IJCSE):
Login with your Inderscience username and password:

Username:        Password:          Forgotten your password?

Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com