Feature Evaluation by Filter, Wrapper, and Embedded Approaches - Feature Selection for Data and Pattern Recognition - page 27

Chapter 3
Feature Evaluation by Filter, Wrapper,
and Embedded Approaches

Urszula Sta nczyk

Abstract The choice of particular variables for construction of a set of characteristic
features relevant to classification can be executed in a kind of external process with
respect to a classification system employed in pattern recognition, it can depend
on the performance of such system, or it can involve some inherent mechanism,
build-in in the system. The three types of approaches correspond to three categories
of methodologies typically exploited in feature selection and reduction: filters, wrap-
pers, and embedded solutions, respectively. They are used when domain knowledge
is unavailable or insufficient for an informed choice, or in order to support this expert
knowledge to achieve higher efficiency, enhanced classification, or reduced sizes of
classifiers. The chapter illustrates the combinations of the three approaches with the
aim of feature evaluation, for binary classification with balanced, for the task of
authorship attribution that belongs with stylometric analysis of texts.

· · · · ·

Keywords Feature evaluation Filter Wrapper Embedded solution DRSA

· ·

ANN Stylometry Authorship attribution

3.1 Introduction

Since inductive learning systems can suffer from both insufficient and excessive
numbers of characteristic features they depend on, the problem of feature selection
and reduction has become quite popular and widely studied, with methodologies
applied typically grouped into three main categories: filters, wrappers, and embedded
solutions [ 17 ].
Filters work independently on a classifier involved in pattern recognition,
regardless of its specifics and parameters [ 14 ]. The choice of attributes is performed
basing on some algorithms, qualitymeasures, for example by referring to information