Performance analysis of GA-based iterative and non-iterative learning approaches for medical domain data sets

摘要

Research in disease diagnosis is a challenging task due to inconsistent, class imbalance, conflicting and high dimensionality nature of medical data sets. The excellent features of each such data set play an important role in improving performance of classifiers that may follow either iterative or non-iterative approach. In the present study, a comparative study is carried out to show the performance of iterative and non-iterative classifiers in combination with genetic algorithm (GA) based feature selection approach over some widely used medical data sets. The experiment assists to identify the clinical data sets for which feature reduction is necessary for improving performance of classifiers. For iterative approaches, two popular classifiers namely C4.5 and RIPPER are chosen, whereas k-NN and Naive Bayes are taken as non-iterative learners. In total, 14 real world medical domain data sets are selected from the University of California, Irvine (UCI repository) for conducting experiments over the learners. From experiments using GA-based feature selection or its absence, it is observed that the naive Bayes provides the best results on most datasets; however, it shows comparatively better performance when features are filtered out.