ACCURACY AND TIME PERIOD COMPARISON USING MODIFIED K-MEAN ALGORITHM

Abstract

The K-mean algorithm is a popular clustering algorithm and has its application in data mining, image segmentation, bioinformatics and many other fields. This algorithm works well with small datasets. In this paper we proposed an algorithm that works well with large datasets. Modified k- mean algorithm avoids getting into locally optimal solution in some degree, and reduces the adoption of cluster -error criterion. The algorithm involves initial centroid selection, which is done randomly in existing Algorithm. Hence it proposes an algorithm which selects initial centroids based on the distances calculated from the origin. One of the most popular clustering algorithms is k- means clustering algorithm.