<![CDATA[Predicting student graduation is a crucial aspect to assess the quality and credibility of higher education institutions. Naive Bayes and K-NN algorithms have been recognised for their effectiveness in predicting graduation. However, most of these studies are limited to academic data. Meanwhile, the variables of thesis completion duration and thesis start time are rarely studied. This study aims to compare the performance of Naive Bayes Gaussian and K-NN algorithms in predicting student graduation using the CRISP-DM method. The data used in this research is the data of students of informatics study programme of Magelang muhammadiyah university. Unlike previous studies that only rely on academic data such as ipk, gender, age, marital status, employment status, and stress level. This research includes the duration of thesis completion and thesis start time as key variables. To compare the performance of Naive Bayes Gaussian and K-NN algorithms, this study adopted three data sharing scenarios: scenario 1 60% training data 40% testing data, scenario 2 70% training data 30% testing data, and scenario 3 80% training data 20% testing data. The results showed that the K-NN algorithm in scenario 2 showed the highest accuracy reaching 91% with precision, recall, and f1-score values of 83.5%, 87.5%, and 85.5%, respectively. On the other hand, Naive Bayes Gaussian reached a maximum accuracy of 88% in Scenario 1 with precision, recall, and f1-score reaching 93%, 77.5%, and 82%, respectively. The research findings show that the K-NN algorithm is superior in predicting student graduation compared to Naive Bayes Gaussian.]]>
Copyrights © 2024
