Optimizing K-Means Initial Number of Cluster Based Heuristic Approach: Literature Review Analysis Perspective

  • Harunur Rosyid
  • Ramlah Mailok
  • Muhammad Modi Lakulu
Keywords: Clustering, Heuristic, K-Means, Number of Cluster

Abstract

Optimizing K-Means Initial Number of Cluster Based Heuristic Approach: Literature Review Analysis Perspective

One popular clustering technique - the K-means widely use in educational scope to clustering and mapping document, data, and user performance in skill. K-means clustering is one of the classical and most widely used clustering algorithms shows its efficiency in many traditional applications its defect appears obviously when the data set to become much more complicated. Based on some research on K-means algorithm shows that Number of a cluster of K-means cannot easily be specified in much real-world application, several algorithms requiring the number of cluster as a parameter cannot be effectively employed. The aim of this paper describes the perspective K-means problems underlying research. Literature analysis of previous studies suggesting that selection of the number of clusters randomly cause problems such as suitable producing globular cluster, less efficient if as the number of cluster grow K-means clustering becomes untenable. From those literature reviews, the heuristic optimization will be approached to solve an initial number of cluster randomly.

Published
2019-12-03
Section
Articles

    Views : 55
    Downloads : 41