THE EFFECT OF FEATURE SELECTION TECHNIQUES ON THE ACCURACY OF HEART DISEASE PREDICTION USING MACHINE LEARNING

Abstract

Artificial intelligence has recently had a significant impact, particularly on the healthcare sector. The use of machine learning has made it possible to predict a number of serious diseases that are now difficult to identify in the medical industry. In this study, the Heart Attack Analysis Prediction Dataset was considered for testing. This dataset was obtained from the Kaggle. The dataset contains 14 features and 303 patient records. To find the best classification algorithm with the highest accuracy, seven feature selection algorithms and eight classification algorithms were used. Simple logistic and Logistic Model Tree classification algorithms were found to be the best classification algorithms for the heart attack analysis and prediction dataset with 85.1485% accuracy. The accuracy of the classification was impacted with the number of features selected.