Prediksi Performansi Mahasiswa dengan Mempertimbangkan Motivasi Intrinsik Menggunakan Machine Learning

Abstract

One of the factors that influence the success of an undergraduate student is learning motivation. Students are required to learn and develop themselves independently and actively find the source of knowledge, not just getting knowledge from the lecturers alone. One type of motivation is intrinsic motivation. Intrinsic motivation can be in the form of satisfaction in undergoing learning and gaining knowledge, the existence of appreciation of the achievements achieved, and the existence of life goals to be achieved. Thus it is necessary to do research to predict student learning performance by considering the intrinsic motivation of students. Primary data was obtained from the motivational questionnaire of 100 students, while secondary data includes attendance data, quiz data, assignments, mid-term exam, and final exam. Furthermore, the two types of data are combined. The first stage uses the Logistics Regression Algorithm, Support Vector Classifier, Decision Tree, Random Forest, Gaussian Naïve Bayes, and K-Nearest Neighbors. The quality of the algorithm is measured using the level of accuracy. Cross validation test with 5 K-Fold was carried out, and the Decision Tree algorithm was obtained with the highest yield of 0.898051. The second stage is to do a tuning hyperparameter using a Grid Search and obtained a value of 0.927206. The third stage is to predict data test as much as 21 data and obtained accuracy of 0.904761.