5-Fold Cross Validation on Supporting K-Nearest Neighbour Accuration of Making Consimilar Symptoms Disease Classification
Andy Satria, Opim Salim Sitompul, Herman Mawengkang
Abstract
Classification is a technique in machine learning that is used to built the group of data. The data that consist of class and target are grouped based on the data attachment to the sample data, so that the data group can be used as a basis for making decisions at a later stage. This study aims to find a classification of two diseases, dengue fever and typhus which have similar symptoms using a combination of two techniques, 5-Fold Cross Validation and K-Nearest Neighbor. The combination of these techniques is done to improve the accuracy of the classification, feature selection is done in preprocessing phase to collect verified feature to get information gain to measure which feature is kept and continue to next stage of classification process. A confusion matrix will be applied to test the performance in terms of the accuracy of the constructed classification. K-Nearest Neighbor (KNN) is a classification method that applies a supervised algorithm, while 5-Fold Cross Validation is a statistical method that is applied to test the performance of a method or technique. This investigation resulted in an increase in the value of accuracy and recall, with a score of 1.79 points and a recall value of 96.43%.