Abstract: Feature selection process plays a vital role in data mining domain, which engrosses recognizing a subset of the good number of practical features that constructs well-matched outcomes as the innovative complete deposit of features. In this paper the algorithm called Feature Selection could be experimented by means of both competence and usefulness. At the same time as the competence apprehensions the time obligatory to come across a subset of features, the usefulness is associated to the eminence of the subset of features. An innovative algorithm called a “Fast Cluster Based Feature Selection (FAST)” is proposed and the experimental results show that FAST not only produces lesser subsets of features but also get better the presentations of the classifiers.

Keywords: Feature Extraction, Subset, Clusters, FAST, Filtering Process.