A REVIEW PAPER ON UNSUPERVISED FEATURE SELECTION ALGORITHMS

Abstract

Enormous amount of data is available and the major challenging task is to find an efficient way of utilizing those data. Due to presence of noise, redundancy and irrelevant data the dimensions of data increases. Feature selection technique aims in selecting a small subset of the significant features from the original features and it’s known as preprocessing step. Feature selection methods are classified as Supervised Feature selection, Unsupervised Feature selection, Semi-supervised Feature selection based on the availability of class labels. In Supervised Feature selection method class label is used for selecting the relevant features where in Unsupervised Feature selection method unlabeled data are taken for consideration. Practically the data sets available now a days are unlabeled and the hence we are in need to find a best unsupervised feature selection method which produces the effective result. This paper analyses some of the existing unsupervised feature selection algorithms.