<![CDATA[The covid-19 pandemic forced students and lecturers to carry out teaching and learning from home. Therefore, XYZ University focuses its students on using e-learning. E-learning that has been running and used by students must be evaluated, so that teaching and learning activities can run well. Evaluation can be done by collecting opinions based on the features of XYZ University E-learning on students through questionnaires. All opinions can be analyzed using classification method called Naïve Bayes and Support Vector Machine for comparison. The research started by collecting data, preprocessing data, labeling using polarity, calculating the frequency that often from each e-learning feature, and calculating the accuracy of the Complement Naïve Bayes model and Support Vector Machine model. The research results conducted on 1995 dataset testing, in student opinions with 1289 positive values, 372 negative values, and 364 neutral values. Reinforced by the comparison result of Complement Naive Bayes and Support Vector Machine. When Complement Naïve Bayes model accuracy of 89%, recall 85,3%, and the f1-score 85%. While Support Vector Machine accuracy is lower 11,1% than Complement Naïve Bayes Model with only 74,4%. These results indicate that of the 12 features on XYZ University E-learning, 8 features have a good opinion, 2 features have a bad opinion, and 2 feature have a neutral opinion.]]>
Copyrights © 2023
