Wrapper method for feature selection to classify cardiac arrhythmia

Efficient monitoring of cardiac patients can save tremendous amount of lives. Cardiac disease prediction and classification has gained utmost significance in this regard during the past few years. This paper presents a predictive model for classification of arrhythmias. The model works by selecting best features using wrapper algorithm around random forest, followed by implementing various machine learning classifiers on the selected features. Cardiac arrhythmia dataset from University of California, Irvine (UCI) machine learning repository has been used for the experimental purpose. After normalizing the data, repeated cross validation with 10 folds is applied on support vector machine (SVM), K nearest neighbor (KNN), Naïve Bayes, random forest, and Multi-Layer perceptron (MLP). The experimental results demonstrate that MLP beats other classifiers by achieving an average accuracy of 78.26%, while accuracies calculated for KNN and SVM are 76.6% and 74.4% respectively, outperforming the accuracies of previous models.