Entropy-Based Classifier Enhancement to Handle Imbalanced Class Problem
Procedia Computer Science 2017
Arnis Kiršners, Sergejs Paršutins, Henrihs Gorskis

The paper presents a possible enhancement of entropy-based classifiers to handle problems, caused by the class imbalance in the original dataset. The proposed method was tested on synthetic data in order to analyse its robustness in the controlled environment with different class proportions. As also the proposed method was tested on the real medical data with imbalanced classes and compared to the original classification algorithm results. The medical field was chosen for testing due to frequent situations with uneven class ratios.


Keywords
Classification, Decision tree, Imbalanced class problem
DOI
10.1016/j.procs.2017.01.176
Hyperlink
http://www.sciencedirect.com/science/article/pii/S1877050917301771

Kiršners, A., Paršutins, S., Gorskis, H. Entropy-Based Classifier Enhancement to Handle Imbalanced Class Problem. Procedia Computer Science, 2017, Vol. 104, pp.586-591. ISSN 1877-0509. Available from: doi:10.1016/j.procs.2017.01.176

Publication language
English (en)
The Scientific Library of the Riga Technical University.
E-mail: uzzinas@rtu.lv; Phone: +371 28399196