Improved Variance K Means Algorithm using Multi Objective Genetic Algorithm for Validate Cluster Generation

In past decade, several methods have been introduced to identify the solutions of multiple clustering. Arrangement of these strategies is chiefly in view of the examination of genuine information space, space nature transformation, and sub-space projections. In this paper, an improved k-means Multi Objective Genetic Algorithm(MOGA) is proposed for detecting a generic optimal separation of the given heterogeneous numeral and categorical data within a clearly identified number of clusters Proposed method integrates the genetic algorithm within the k-means algorithm with improved cost function to manage the numeral data. For the effective evaluation of the proposed algorithm three original datasets are used from UCI largest dataset repository center. Experimental result shows the effectiveness of the proposed algorithm in regaining the unexpressed cluster designs from categorical dataset if such designs alive. Improved illustration for cluster center is used which can draw cluster behavior with effectiveness because it carries the distribution of all the unreserved values in Cluster. Comparative analysis showed the superiority of proposed algorithm over VK-means algorithm.