Classification of Heart Disease Using Linear Discriminant Analysis Algorithm

Abstract

Ischaemic coronary heart disease is the number one cause of death globally. Detecting this disease can only be done by consulting directly with a cardiologist at a cost that is certainly not small. Therefore, is a need for a system to detect heart disease in patients with accuracy but low cost. With the development of technology, especially in artificial intelligence area, there was machine learning techniques to enhance automatic detection capabilities. Linear Discriminant Analysis are one of machine learning method for prediction to detect heart disease as early as possible. In this study, linear discriminant analysis algorithm was implemented to classify heart disease. Dataset used are from the UCI machine learning repository. This study carried out two experimental conditions, classifying heart disease based on suffer or not, other is classifying heart disease by 5 level stage. Result proves that the performance of the classifier with LDA with 2 classes is better than 5 classes. Performance of the LDA algorithm in classifying heart disease with 2 labels that are used as targets or outputs. From these results, the precision value is 0.82, the recall value is 0.81, the F1 score value is 0.81, with an accuracy of 81.22%.