An Improved Fast Clustering method for Feature Subset Selection on High-Dimensional Data clustering

J.K.MADHAVI , G.VENKATESH YADAV

ABSTRACT In this paper, we proposed Feature extraction as the process of eliminating the irrelevant information and features during Data Mining. Feature subset selection can be analyzed as the practice of identifying and removing as lot of inappropriate and unnecessary features as achievable. This if for the reason that, irrelevant features do not contribute to the predictive accuracy and redundant features do not redound to receiving a better analysis for that they provide typically information which is previously present in other features of all the existing feature subset selection algorithms, most of them can effectively eliminate irrelevant features but fail to handle redundant features. The improved FAST algorithm is evaluated using various types of data like text data, micro-array data and image data to represent its performance. Fast clustering algorithm work can be done in two steps. The first step is to moving out irrelevant features from the dataset, for irrelevant features are removed by the features having the value above the predefined threshold. And the second step is to eliminate the redundant features from the dataset, the redundant features is removed by constructing the Minimum Spanning Tree and separate the tree having the edge distance more than its neighbor to form the separate clusters, from the clusters features that are strongly associated with the target features are selected to form the subset of features. The Fast clustering Algorithm is more efficient than the existing feature subset selection algorithms. These can be formed in well equipped format and the time taken to retrieve the information will be short time and the Fast algorithm calculates the retrieval time of the data from the dataset. This algorithm formulates as per the data available in the dataset. By analyzing the efficiency of the proposed work and existing work, the time taken to retrieve the data will be better in the proposed by removing all the irrelevant features which gets analyzed. Keywords:- Irrelevant and redundant features, fast clustering-based feature selection algorithm, feature subset selection, Data Mining, Filter method, featured clustering, Data search, Text classification, Clustering, Rule mining

Title:	An Improved Fast Clustering method for Feature Subset Selection on High-Dimensional Data clustering
Author Name:	J.K.MADHAVI , G.VENKATESH YADAV
Abstract:	ABSTRACT In this paper, we proposed Feature extraction as the process of eliminating the irrelevant information and features during Data Mining. Feature subset selection can be analyzed as the practice of identifying and removing as lot of inappropriate and unnecessary features as achievable. This if for the reason that, irrelevant features do not contribute to the predictive accuracy and redundant features do not redound to receiving a better analysis for that they provide typically information which is previously present in other features of all the existing feature subset selection algorithms, most of them can effectively eliminate irrelevant features but fail to handle redundant features. The improved FAST algorithm is evaluated using various types of data like text data, micro-array data and image data to represent its performance. Fast clustering algorithm work can be done in two steps. The first step is to moving out irrelevant features from the dataset, for irrelevant features are removed by the features having the value above the predefined threshold. And the second step is to eliminate the redundant features from the dataset, the redundant features is removed by constructing the Minimum Spanning Tree and separate the tree having the edge distance more than its neighbor to form the separate clusters, from the clusters features that are strongly associated with the target features are selected to form the subset of features. The Fast clustering Algorithm is more efficient than the existing feature subset selection algorithms. These can be formed in well equipped format and the time taken to retrieve the information will be short time and the Fast algorithm calculates the retrieval time of the data from the dataset. This algorithm formulates as per the data available in the dataset. By analyzing the efficiency of the proposed work and existing work, the time taken to retrieve the data will be better in the proposed by removing all the irrelevant features which gets analyzed. Keywords:- Irrelevant and redundant features, fast clustering-based feature selection algorithm, feature subset selection, Data Mining, Filter method, featured clustering, Data search, Text classification, Clustering, Rule mining
Cite this article:	J.K.MADHAVI , G.VENKATESH YADAV , " An Improved Fast Clustering method for Feature Subset Selection on High-Dimensional Data clustering " , International Journal of Application or Innovation in Engineering & Management (IJAIEM), Volume 3, Issue 10, October 2014 , pp. 026-030 , ISSN 2319 - 4847.
Full Text [PDF] Home