An ensemble based on distances for a kNN method for heart disease diagnosis

This paper introduces an ensemble based on distances for a kNN (k Nearest Neighbor) method and shows results of its application to heart disease diagnosis. The ensemble has been implemented with two configurations. One using three distances and another one using five. We also added to them a weighted version based on the average accuracy that each distance gives when used in the kNN method. Our ensemble gave an average accuracy of nearly 85% for any of the configurations and versions we tested with the UCI heart disease Cleveland data set.