MEASURING THE PERFORMANCE OF THE NAÏVE BAYES ALGORITHM IN DETERMINING STUDENTS' ABILITY IN THE ITSJava APPLICATION

Abstract

The ITSJava application is used online to increase students' understanding and interest in PBO learning. One of the components of the ITSJava application is the domain/expert model. The expert domain model in the ITS Java application determines whether students can continue to the following material or repeat the material being studied. This research aims to determine the accuracy, precision and sensitivity of the results provided by the Naïve Bayes algorithm in determining the level of students' ability to understand the material in the ITSJava application. The research method uses the Knowledge Discovery in Database (KDD) design, which consists of 1) Data selection, 2) Preprocessing, 3) Transformation, 4) Datamining, and 5) Interpretation/evaluation. This research obtained an accuracy level of 89%, precision = 0.92 and recall = 0.73 if using cross-validation with fold = 10, and an accuracy level of 75%, precision = 1, and recall = 0.37 if using percentage split = 80. These results show that the Naïve Bayes method applied to the ITS Java application can determine the level of students' ability to understand the material.