Application of SMOTE Method on Topic Based Question Classification Using Naïve Bayes Algorithm

Abstract

In today's digital era, the utilization of technology in education is essential to support the learning process. This research discusses the classification of junior high school mathematics questions using the Naïve Bayes method. The use of an automated system in question classification helps reduce time and effort in grouping questions based on topics. The Naïve Bayes method was chosen because of its simplicity and ability to process data. The results showed that Naïve Bayes with SMOTE and Math symbols achieved 69% accuracy, while without SMOTE, the accuracy was lower. Cross-validation showed that the classification without symbols attained an accuracy of 89.35%, slightly superior to the classification using symbols, which was 88.79%. This result indicates that Naïve Bayes with SMOTE is more effective. Although the difference in accuracy with or without symbols is slight 0.56%, the performance is relatively equivalent, with an accuracy of 89%.