Big data is one of the most discussed topic along with the development of technology. As a result, many methods have been developed and used in order to extract as many useful information as possible from complex data sets. The versatile use of these techniques also bring a lot of interest in using them in various fields, one of them being educational field. This thesis involves applying these data mining techniques to data sets collected in Universitas Pelita Harapan to create models for predicting students final grade in Calculus and Statistics courses taken by first and second year students. The techniques used are Naïve Bayes, C5.0 Decision Tree, Classification and Regression Tree (CART), and Support Vector Machine (SVM). Models produced by these methods are summarized and analyzed. The results indicate that while SVM and decision tree models have very similar performances, decision tree models are more interpretable. Naïve Bayes with no Laplace estimator perform well compared to when Laplace estimator is used, producing generally higher test accuracy and kappa values. Aside from exams, quizzes, and assignments, the result shows that students’ major, batch, confidence and satisfaction in classes taken may affect the students’ performance in the courses.