HIGH DIMENSIONAL UNSUPERVISED CLUSTERING BASED FEATURE SELECTION ALGORITHM

Feature selection is a process which selects the subset of attributes from the original dataset by removing the irrelevant and redundant attribute. Clustering is the technique in data mining which group the similar object in to one cluster and dissimilar object into other cluster. Some clustering technique does not support high dimensional dataset. By applying the feature selection as a preprocessing step for the clustering make it possible to handle the high dimensional dataset. Feature selection reduce the computational time greatly due to reduced feature subset and also improve clustering quality. Feature selection methods are available for supervised and unsupervised learning. This paper is related to working of feature selection method which is applied on different feature selection algorithm. The result proved that Feature selection through feature clustering algorithm is reduced the more attributes than the standard feature selection algorithm like relief and fisher filter.