Reduct Computation Through Ranking for Efficient Feature Extraction

Data mining process suffers from the curse of dimensionality. If a dataset has a large number of attributes, then the corresponding classification procedure loses its accuracy and precision. In this paper, an efficient algorithm for dimensionality reduction using Reducts through Ranking based on Rough Set Theory is presented. Classification process is accelerated by using efficient reduct for datasets. Reducts resemble the set of attributes which are unique to the data object and can distinguish many objects from each other. A ranking procedure ranks the attributes according to their ability to contribute to the classification procedure. A threshold limit is used to effectively produce efficient Reducts.