<![CDATA[Abstract. YouTube has become one of the most widely used social media platforms for the public to express opinions on various issues, including the policy restricting social media access for children under the age of 16. The large volume of comments generated necessitates sentiment analysis based on machine learning to ensure a more effective classification process. This study aims to analyze the distribution of public sentiment and compare the performance of the Naive Bayes, Support Vector Machine (SVM), and K-Nearest Neighbor (K-NN) algorithms. This research employs a quantitative approach using the CRISP-DM methodology. The data were collected through crawling YouTube comments, followed by preprocessing, TF-IDF weighting, split validation, and classification using the three algorithms. The results show that out of 2,646 comments, there are 811 positive sentiments, 622 neutral sentiments, and 1,213 negative sentiments, indicating that negative sentiment is the dominant category. Based on the evaluation results, the SVM algorithm demonstrates the best performance with an accuracy of 72.08%, precision of 71.85%, recall of 72.08%, and an F1-score of 71.30%, outperforming Naive Bayes and K-NN. These findings indicate that SVM is the most effective algorithm for classifying sentiment in YouTube comments on this topic]]>
Copyrights © 2026
