Feature Selection Techniques with Distributed Data Mining Models

Data mediated knowledge discovery is essential for any end users for value added decision making. Discerning vital, accurate and precise knowledge in the classification, various feature subsets are necessary. Apart from feature selection, processing and representation of data is also indispensable for analysis and implementation of any knowledge. Principal Component Analysis is the used for data pre-processing and representation of data. Eigen vectors, co variance matrix are estimated for distributed environment where local and global set are computed and evaluated. It reduces the dimensionality of data. RELIEF, CMIM and other feature selection methods are discussed here in this paper. On selecting the features may increase the classification accuracy and enhance classification and prediction.