Determining single tuition fee of higher education in Indonesia: A comparative analysis of data mining classification algorithms

Abstract

Student's Single Tuition Fee or Uang Kuliah Tunggal (UKT) is a subsidy policy in higher education by the Indonesian government. This policy regulates the tuition fees incurred by each student at each semester in every higher education institutions. Since the cost of UKT expenses is influenced by the financial ability of each student, therefore the cost of education among students must be grouped into several classes. Until recently, there has been no standard to make such classification whereas such determination is an important task to solve by every higher institution in Indonesia. This study aims to compare five data mining classification algorithms (Gaussian Naïve Bayes, Multinomial Naïve Bayes, Bernoulli Naïve Bayes, Decision Tree and SVM) to find the best algorithm for the case of determining the UKT classes. The experiment is conducted using 230 training data and 10-fold cross-validation evaluation. Based on the result, Decision Tree managed to obtain average accuracy value of 0.814 or 81.4%. Finally, Decision Tree is used to classify the UKT classes of3258 data of students.