Original scientific paper

https://doi.org/10.31341/jios.43.1.1

Discretization for Naive Bayes Taking the Specifics of Heart Data into Account

Jan Bohacik orcid.org/0000-0001-7407-4502 ; Faculty of Management Science and Informatics, University of Zilina, Zilina, Slovakia
Michal Zabovsky ; University Science Park, University of Zilina, Zilina, Slovakia

Abstract

At the present time heart disease is a major cause of death. Factors such as physical inactiveness, obesity, diabetes, social isolation and aging are expected to make the situation worse. It is worsened even further with misdiagnosis of patients describing heart related issues. A probability decision support approach to diagnosis of heart disease based on Naive Bayes is discussed here as most hospitals collect patient records but these are rarely used for automatic decision support. The approach is analyzed on Statlog heart data with the focus on improving preprocessing methods. As the result, a discretization algorithm with Equal Frequency Discretization which considers the specifics of engaged heart disease patients is presented. Enhancements of achieved accuracy with the added discretization and in comparison with other machine learning algorithms are shown in experiments founded on 10-fold cross-validation.

Keywords

discretization; Naive Bayes; diagnosis; heart disease

Hrčak ID:

221383

URI

https://hrcak.srce.hr/221383

Publication date:

21.6.2019.

Visits: 1.282 *