Adobe PDF (1.64 MB)
Title Details:
Alternative Methods and Special Topics in Classification
Authors: Kirkos, Efstathios
Reviewer: Symeonidis, Panagiotis
Description:
Abstract:
A wide variety of methods can be used for classification. This chapter provides a concise yet substantive description of these methods. The reader will understand their characteristics to apply them in practice, compare them, and ultimately select the most appropriate approach. The chapter covers classification with Multilayer Perceptron Neural Networks, Bayesian Networks, instance-based classifiers, and Support Vector Machines (SVMs). For each method, the advantages and disadvantages are presented. The chapter also discusses complex classifiers, including hybrid and ensemble approaches. Special classification issues are addressed, such as class imbalance and differential error costs. Methods for evaluating classifiers are presented, including Holdout, Cross-Validation, and ROC curves.
Technical Editors: Papavasileiou, Spyridon
Type: Chapter
Creation Date: 2015
Item Details:
License: http://creativecommons.org/licenses/by-nc-nd/3.0/gr
Handle http://hdl.handle.net/11419/1237
Bibliographic Reference: Kirkos, E. (2015). Alternative Methods and Special Topics in Classification [Chapter]. In Kirkos, E. 2015. Business Intelligence and Data Mining [Undergraduate textbook]. Kallipos, Open Academic Editions. https://hdl.handle.net/11419/1237
Language: Greek
Is Part of: Business Intelligence and Data Mining
Publication Origin: Kallipos, Open Academic Editions