Comparative Study of Feature Subset Selection Methods for Dimensionality Reduction on Scientific Data

Feature Selection(FS) is an important step to enhance the classification accuracy. Using the lazy learning classification algorithm, the feature selection methods calculate relevancy to reduce the storage. Dimensionality reduction technique on scientific data is a popular area to understand the underlying scientific knowledge in a data set, resulted from scientific experiments. This paper presents a review and systematic comparative study of methods and techniques used in scientific data mining. The performances of the techniques are compared and a meaningful direction has been arrived. It is understood that there are several techniques such as Sequential Forward Selection(SFS), Sequential Floating Forward Selection(SFFS) and Random Subset Feature Selection (RSFS) etc., which are used to minimize the storage space of the scientific data set in combination of Nearest Neighbor classifier. The paper explains these approaches by identifying various dimensionality reduction techniques to improve the performance.