Author(s): S. M. A. Khaleelur Rahman, M. Mohamed Sathik, K. Senthamarai Kannan
Data mining techniques have been used for finding useful patterns in data. Sometimes these patterns are visible easily. Many times, there are lack of patterns or very excessive patterns. Real world d ata usually contains more complex structures so that predicting useful pattern is not easy for even a sophisticated data mining technique. Clustering provides a method to break a larger database in to meaningful pieces to describe each piece simply. Once the proper clusters have been identified and defined then it is possible to find patterns within each cluster. In this paper , a proposed method based on variable clustering method is presented. We first perform VARCHA approach. This is a bottom up hierarchical agglomerative approach relies on popular K-MEANS algorithm. A top down method VARCLUS is also used .Choosing right number of cluster is an important criterion in clustering problems. Required number of clusters is decided by calculating Variation and Proportion. Experimental results clearly indicated that our methods are effective for choosing number of clusters. The test results suggested our proposed method can be applied to different data sets for effective results.