Analisis Sentimen Siswa Terhadap Pelajaran Informatika di SMPK St. Yohanes Kalembu Lona Dengan Metode Naive Bayes Classifer

Abstract

With the increasingly rapid development of science and technology today, the learning process related to information and communication technology is not only studied in universities but has been taught to students in junior secondary schools (SLTP) and senior secondary schools (SLTA), with There is a hope that Information and Communication Technology (ICT) subjects can contribute to the realization of critical and creative reasoning power. Since its implementation in the Education Unit Level Curriculum (KTSP), ICT subjects have become one of the subjects whose grades are also used as criteria for students' grade promotion. In this research on student sentiment toward informatics lessons, positive, negative, and neutral opinions will be grouped using the Naïve Bayes Classifier method. The research process carried out by the author in this study was carried out by collecting data, labeling the data set, and pre-processing the dataset which consisted of cleaning, case folding, tokenizing, stopword removal, stemming, and then continued with the TF-IDF calculation process. The aim of applying the Na????̈ve Bayes Classifier in this research is to group sentiments which are then followed by system validation and evaluation using the K-Fold Cross Validation and Confusion Matrix methods. The data set used is the opinion of students at SMPK St. Yohanes Kalembu Lona collected using a questionnaire that was distributed randomly to students in Class VII, VIII, and Class IX with 5 questions related to informatics lessons to each of the 40 students selected at random. The level of accuracy produced by the Na????̈ve Bayes Classifier method is based on test results using the splitting dataset technique, the results are adapted to the specified percentage, namely 80% compared to 20%, the results obtained are 85%, while testing using sub-datasets produces results of 84%, 83% respectively. , 80%, 82%, and 84% for the 5 sub-datasets while for the level of accuracy using the K-fold Cross Validation technique the results obtained for each dataset are 84%, 83%, 82%, 78%, and 80%.