Comparison of Classification Algorithms for Predicting Graduation of Informatics Engineering Students with Orange Data Mining

Timely completion of studies is an important parameter to assess the competence of graduates. However, challenges arise because not all students can complete their studies according to the predetermined schedule. In this study, the prediction of student graduation status was developed using four classification models, namely Decision Tree, Naïve Bayes, K-NN and SVM. This research also involves the addition of new variables to improve the accuracy of predicting student graduation rates. The dataset used is the 2018-2020 batch student data at Prof. Dr. Hamka Muhammadiyah University, consisting of 500 student data (60% training data and 40% test data). The analysis was carried out by utilizing Orange Data Mining software, with model evaluation involving K-Fold Cross Validation with a value of (K=5), Confusion Matrix, and ROC. The results show that the K-NN algorithm is the most effective algorithm in predicting student graduation status, with accuracy and precision rates reaching 92%, recall rate of 89%.