Author(s): S.I.S.Jaffarvalli, G.Surya Narayana

In terms of high dimensional data a cluster is defined is a connected region of multi dimensional space containing a relative high density of points separated from other such regions contains a relative low density of points. Cluster analysis is an unsupervised learning. The main aim of this paper is to remove irrelevant and redundant features and increase the level of accuracy. In the proposed Feature selection and extraction is the special form of dimensionality decline where feature selection is the subfield of feature extraction mining. The efficiency concerns related to time for finding subset of features and the effectiveness is related to the quality of the subset selections. To ensure the quality and efficiency of FAST, the efficient minimum spanning tree algorithm called MST clustering method is implemented. The proposed extensive experiments are carried out to compare FAST algorithm and several representative feature subset selection algorithms, namely, ReliefF, FCBF, CFS, FOCUS and Consist with respect to various types of well-known classifiers.