An essential component of an intelligent data analysis process is clustering, an unsupervised learning technique. By grouping the patterns into homogeneous clusters, it facilitates the investigation of the links between the patterns. In the realm of information retrieval(IR), clustering has been dynamically applied to arrange of applications. One of the most active areas of study and development nowadays is clustering. By using clustering, one can find the set of significant groups in which members are more linked to each other than to members of other groups. There salting groupings can offer a framework for arranging length by text sections to make browsing and searching easier. Numerous clustering techniques have been thoroughly examined in relation to the clustering problem. Expectation Maximization(EM) and its variations, as well as the well-known link-means algorithm, are examples of iterative optimization clustering algorithms that have been shown to perform rather well for clustering. These algorithms are still among the most popular and effective. In the heart spect dataset, which has the following features: purity, entropy, CPU time, cluster-wise analysis, mean value analysis, and inter-cluster distance, this study examines the partition method clustering approaches, EM, Kmeans and K*Means algorithm. In order to support the conclusion that the behaviour in clusters produced by the EM algorithm is of a higher calibre than that of the k-means and k*means algorithms, the research finally presents the experimental results from datasets for five clusters.