ANALYSIS THE CLUSTER PERFORMANCE OF REAL DATASET USING SPSS TOOL WITH K-MEANS APPROACH VIA PCA

Abstract

Partitioning problems are handled by the idea of cluster and this technique which plays the essential work in mining of data from the given dataset. The K-Means cluster is well accepted theory to apply on huge datasets, but has some drawbacks. The factual dataset is taken from the repository of data used for clustering. Furthermore, as getting the outcome of this procedure is essential to resolve the limitations and quality enhanced of cluster by apply the Principal Component Analysis (PCA) on the dataset. In paper we have demonstrate the results by experimental for factual datasets with dissimilarities. We have worked to validate the experimental significant for the clusters metric and component size minimized for different dataset during the processing on SPSS tool on the basis of eigenvalues. In this research paper we also discussed the comparative analysis of distance between initial centroid of wine and disease of heart dataset at the level of cluster k=2 and k=3.