A Comparison of Rule Induction Using Feature Selection and the LEM2 Algorithm - Feature Selection for Data and Pattern Recognition - page 157

Chapter 8
A Comparison of Rule Induction
Using Feature Selection and the LEM2
Algorithm

Jerzy W. Grzymała-Busse

Abstract Themain objective of this chapter is to compare a strategy of rule induction
based on feature selection, exemplified by the LEM1 algorithm, with another strategy,
not using feature selection, exemplified by theLEM2 algorithm. TheLEM2 algorithm
uses all possible attribute-value pairs as the search space. It is shown that LEM2
significantly outperforms LEM1, a strategy based on feature selection in terms of
an error rate (5% significance level, two-tailed test). At the same time, the LEM2
algorithm induces smaller rule sets with the smaller total number of conditions as
well. The time complexity for both algorithms is the same.

· · ·

Keywords Rough set theory Feature selection LERS datamining system LEM1

and LEM2 rule induction algorithms

8.1 Introduction

In 1982 an approach to feature selection, under the name of attribute reduction, using
rough set theory, was introduced in [ 26 ], see also [ 27 , 28 ]. In the rough set community
reducing the original attribute set of attributes is one of the main and frequently used
techniques.
Feature selection is the process of selecting a subset of relevant features. Research
on feature selection, see, e.g., [ 2 , 6 , 20 - 23 , 29 , 31 ], includes finding the smallest set
of features, improving this way the efficiency of data processing. Data are presented
in tables, with rows labeled as cases (examples or entries) and columns labeled as
features (variables or attributes).