The Philosophy of Naive Bayes and its Comparison with the Tree Augmented Naive Bayes (TAN) in Making Predictions (Case Study Using Course Student Data)

Abstract

This paper aims to present a comparison between the Naive Bayes model and the Tree Augmented Naive Bayes (TAN). This comparison is preceded by a philosophical explanation of the two in the author's way of understanding. Then this comparison was carried out in the case of making a predictive model for the final exam scores of students at the Muhammadiyah University of Ponorogo. In this case, we apply two different models, namely Naive Bayes and Tree Augmented Naive Bayes (TAN) to predict learning outcomes before ending at the end of the semester in terms of lecturers' assessments of students. When the course is in progress, the lecturer needs to continuously evaluate students' understanding of the subject matter being taught. This is so that lecturers can immediately anticipate learning problems in class. Calculations with these two models use only R language with jupyter notebook interface. Validation and testing of the two models used a case dataset in 4 classes of language theory and automata course students even semester 2017-2018 at the Department of Informatics Engineering, Faculty of Engineering, and University of Muhammadiyah Ponorogo with a dataset size of 99 notes (99 students). For the validation and model testing methods, k- fold and hold-out cross-validation are used. Each model is validated and tested with the same k-fold method scheme.