Sentiment Analysis Of Lecturer's Learning Based On Student's Criticism Data Using Naive Bayes Method

Abstract

The quality of lecturers' teaching is the main key to the success of education at the University. By having lecturers who can provide good teaching, the seeds of nation-changing students can be created. To get lecturers who meet academic standards, of course, evaluation needs to be done. Therefore, Budi Luhur University always conducts questionnaires containing suggestion criticism forms that can be filled in by students. Completion of the suggestion criticism form is carried out for each group of courses at the end of the semester. Currently, the feedback data has not been used to analyze and evaluate the learning process. Therefore, in this research, sentiment analysis is carried out on the results of criticism suggestions that have been sent by students, to find out whether the criticism suggestions are positive, negative, or neutral. One of the sentiment analysis methods that can be used to solve opinion mining problems is the Naive Bayes Method. Data collected as many as 10,067 in the span of 1 semester, namely the odd semester of the 2021/2022 academic year. This suggestion criticism data is then preprocessed, and classified using the Naive Bayes method, testing is carried out using the Naive Bayes Classifier program which is made in the PHP programming language, then accuracy is obtained with the Naïve Bayes method in testing 60% - 40% getting an accuracy result of 83.92%, then in split data 70% - 30% getting an accuracy result of 83.26%, then for split data results 80% - 20% getting an accuracy result of 81.96%, obtained positive sentiment results as much as 2484, then negative sentiment as much as 152, and neutral sentiment as much as 267. This research aims to get sentiment results which are then expected to be used as a reference to improve the quality of teaching lecturers at the university and further evaluation.