Naive Bayes using to predict students' academic performance at faculty of literature

Abstract

In Indonesia, in order to maximize its academic potential, high school students need to be grouped in classes based on their interests and talents. Three commonly made groups are natural science, social science, and linguistics. Problems can arise in the future when students experience a change of interest. One of the cases is that there are students who previously belonged to the group of natural science classes interested in continuing studies in higher education in the field of language and literature. This study aims to assist students with such cases by predicting the likelihood of their success adapting to new environments. Predictions are based on input data on students' activities and skills related to the language field. The Naive Bayes method is used with the input of a number of attributes, including the national exam score of Indonesian and English language, the average of the national exam score, the presence or absence of language-related achievements, and the number of books read each month. Converted GPAs in ordinal form are selected as outputs in the case of this prediction. The results shows that the accuracy of this technique reaches 70 percent, so it can be interpreted that the Naive Bayes method has the potential to answer the question of whether a student can adapt and perform well while studying in language and literature faculty.