A Survey on Fast Clustering Based Feature Selection Algorithm for High Dimensional Data

Data mining has main problem of partitioning a group of objects into a number of subsets, such that similarity in each subset or cluster is increased and effective result should be obtained. The feature selection method is more generalized form of feature extraction. Feature selection gives useful feature from data while the feature extraction creates new feature set according to existing feature sets. The main concept of this fast clustering feature selection algorithm is to cluster subset with most similar characteristic while removing irrelevant subset from that cluster. This FAST algorithm concern with both efficiency and accuracy for finding required set. The FAST algorithm requires two steps for its working, in first step algorithm uses minimum spanning tree (MST) to divide data into different clusters and in second step, it removes irrelevant sets and gives accurate and efficient result with similar sets. All clusters resulted in FAST algorithms are relatively independent of each other. So this may useful for most effective results.