Feature selection is used to improve the efficiency of learning algorithms
by finding an optimal subset of features. However, most feature selection t
echniques can handle only certain types of data. Additional limitations of
existing methods include intensive computational requirements and inability
to identify redundant variables. In this paper, we present a novel, inform
ation-theoretic algorithm for feature selection, which finds an optimal set
of attributes by removing both irrelevant and redundant features. The algo
rithm has a polynomial computational complexity and is applicable to datase
ts of a mixed nature. The method performance is evaluated on several benchm
ark datasets by using a standard classifier (C4.5). (C) 2001 Elsevier Scien
ce B.V. All rights reserved.