A Comparative Study of Student Satisfaction Levels on Online Learning Using K-NN and Naïve Bayes

Abstract

The outbreak of the Covid-19 pandemic in Indonesia led to restrictions on human social activities to minimize transmission. Teaching-learning is also affected when students must stay home and follow distance learning based on Government Regulation Number 21 of 2020, the Large-Scale Social Restrictions (PSBB) policy, issued on March 31, 2020. This has led to the emergence of learning support applications such as Zoom, Google Classroom, Google Meet, E-Learning, and many more. However, this new learning culture requires adaptation for effective implementation. During the adaptation process, researchers want to measure the level of student satisfaction and find out the best algorithm for classifying the level of student satisfaction. This measurement uses two data mining algorithms, K-Nearest Neighbour (K-NN) and Naïve Bayes, and the Islamic State University of Sultan Syarif Kasim Riau students as the research object. Different algorithms have varying strengths and weaknesses in handling specific data types and classification tasks. By comparing both algorithms, we can assess their generalization capabilities. A model that performs well on training data but fails to generalize to unseen data may not be as effective as a more robust algorithm that exhibits better generalization performance. K-NN classification with a value of k = 3 gets good results. Based on the study results, the conclusion is that K-NN is more optimal in classifying student satisfaction levels than Naïve Bayes with an accuracy ratio of 85% : 80%, precision of 85% : 84%, and recall of 99% : 93%.