Performance Evaluation of EM and K-Means Clustering Algorithms in Data Mining System

Abstract

In the Emerging field of Data Mining System there are different techniques namely Clustering, Prediction, Classification, and Association etc. Clustering technique performs by dividing the particular data set into associated groups such that every group does not have anything in common.Clustering algorithms have emerged as an alternative powerful meta-learning tool to accurately analyze the massive volume of data generated by modern applications. Actually the main goal is to classify data into clusters such that objects are clustered in the same cluster when they are related according to particular metrics. Classification is the organization of data sets into some predefined sets using various mathematical models. This research discusses the comparison of algorithms K-Means and Expectation-Maximization in clustering. Empirically, we focused on wide experiments where wecompared the best typical algorithm from each of the categories using a large number of real or bigdata sets. The effectiveness of the Expectation-Maximization clustering algorithm is measured through a number of internaland external validity metrics, stability, runtime and scalability tests.