2010 | OriginalPaper | Chapter
Dimensionality Reduction and Optimum Feature Selection in Designing Efficient Classifiers
Authors : A. K. Das, J. Sil
Published in: Swarm, Evolutionary, and Memetic Computing
Publisher: Springer Berlin Heidelberg
Activate our intelligent search to find suitable subject content or patents.
Select sections of text to find matching patents with Artificial Intelligence. powered by
Select sections of text to find additional relevant content using AI-assisted search. powered by
In the course of day-to-day work, huge volumes of data sets constantly grow accumulating a large number of features, but lack completeness and have relatively low information density. Dimensionality reduction and feature selection are the core issues in handling such data sets and more specifically, discovering relationships in data. Dimensionality reduction by reduct generation is an important aspect of classification where reduced attribute set has the same classification power as the entire set of attributes of an information system. In the paper, multiple reducts are generated integrating the concept of rough set theory (RST) and relational algebra operations. As a next step, the attributes of the reducts, which are relatively better associated and have stronger classification power, are selected to generate the single reduct using classical Apriori algorithm. Different classifiers are built using the single reduct and accuracies are compared to measure the effectiveness of the proposed method.