C4.5 Algorithm Implementation to Predict Student Satisfaction Level of Lecturer’s Performance in the Covid-19 Pandemic

Abstract

Implementation of education during the emergency period of Covid-19 in Higher Education was carried out at home through online/distance learning. The lecturer is one of the key holders of success in the learning process. Lecturer performance is a main factor needed to improve education and service quality in online learning. In this study, the authors implemented the C4.5 algorithm using RapidMiner 9.10 app to predict student satisfaction with lecturer performance during the Covid-19 pandemic. The data in this study were obtained from a questionnaire distributed to active students in the Computer Science Study Program (class of 2016 - 2021) at the University of Nusa Cendana with 942 records. The attributes used in this study were the lecturer's age, gender, suitability of learning media (SLM), and the competencies of Pedagogic Competence (PeC), Professional Competence (PrC), Personal Competence (PsC), and social competence (SC), with the level of student satisfaction as the target class divided into two, namely Satisfied and Dissatisfied. The dataset is processed using RapidMiner and produces 11 decision rules which show that the attribute PeC has the most significant influence on the level of student satisfaction with lecturer performance during the Covid-19 pandemic and the test results of the decision tree model using cross-validation. The test results show that the C4.5 algorithm has a good performance in predicting levels of student satisfaction with an accuracy rate of 94.8%, precision for the prediction class Dissatisfied and Satisfied of 92.23 % and 95.52%, and recall of the actual Dissatisfied and Satisfied classes of 85.2% and 97.77%.