Comparative Study of Soft Computing Techniques on Medical Datasets

Mangesh Metkari, M.A. Pradhan

Abstract: Data classification is process of dividing dataset into two or more different classes where each class contains similar type of data items. In this paper we compare the different classification technique using the WEKA tool that will be helpful for decision making in medical diagnosis. WEKA is open source tool providing classification using soft computing technique for data mining process. Our goal is to analysis of the performance of different classifiers on different medical datasets. The analysis is done for five different medical datasets with four different classifiers in terms of the execution time, correctly classified, incorrectly classified and the mean absolute error. From the obtained results of classifiers we conclude that KNN is the effective classifier for medical dataset than other classifiers we used for the analysis.

Keywords: Random Forest, KNN, Multilayer Perceptrons, Classifier