Effect of the Clonal Selection Algorithm on Classifiers

Abstract

To be able to make the classification process, there should be a sufficient number of samples. Collecting a sufficient number of samples, especially for those dealing with medical data, is a laborious task. To obtain the approval of the ethics committee in our country, patient data coming from a certain time interval rather than a sample number can be requested. Therefore, there are difficulties in reaching a sufficient number of samples. In this study, the effect of the clonal selection algorithm which is one of the artificial immune system algorithms on standard classifiers was investigated. The chronic kidney disease dataset from the university of California Irvine machine learning repository was chosen as the dataset. Among the commonly used methods for classification, methods of k nearest neighbor, decision trees and artificial neural networks were selected as classifiers. While k nearest neighbor is a distance-based algorithm, a decision tree is a regression-based method and the artificial neural network which is quite popular nowadays is a nature-inspired method. According to the results of the experiments, it is found that the data reproduction process by using the clonal selection algorithm has increased the performances of the classifiers.