ANALYSIS OF CLUSTERING IN K-MEANS AND K-MEDOIDS

Abstract

Cluster analysis has long been used in a wide variety of fields, psychology and other social sciences, biology, statistics, pattern recognition, information retrieval,
machine learning, and data mining. Cluster analysis divides data into meaningful or useful groups. Data mining adds to clustering the complications of very large
datasets with very many attributes of different types. There are various types of algorithmsin data mining process. Clustering has taken its roots from algorithms like k- means and k medoids. From these algorithm k-means algorithmis evolved. K-means is very popular because it is conceptually simple and is computationally fast and memory efficient but thereare various types of limitations in k means algorithm that makes extraction somewhat difficult. K- medoids clustering algorithm suffers from many limitations. In this paper we are discussing these merits and demerits in k- means and k-medoidshow to overcome