Sentimen Analisis Komentar Youtube Mengenai Isu Kesehatan Mental di Indonesia Menggunakan Algoritma Naive Bayes

Kesehatan mental merupakan isu yang semakin mendapat perhatian di Indonesia. Meningkatnya penggunaan media sosial, khususnya YouTube, mendorong masyarakat menyampaikan pendapat dan pengalaman mengenai kesehatan mental melalui kolom komentar. Komentar tersebut dapat dimanfaatkan untuk mengetahui kecenderungan sentimen masyarakat menggunakan teknik analisis sentimen. Penelitian ini bertujuan menganalisis sentimen komentar YouTube mengenai isu kesehatan mental di Indonesia menggunakan algoritma Naïve Bayes serta menentukan skenario pembagian data yang menghasilkan kinerja terbaik.
Penelitian menggunakan 900 komentar dari tiga video YouTube bertema kesehatan mental yang diperoleh melalui proses scraping. Data diseleksi dan diberi label secara manual oleh tiga annotator menggunakan metode majority voting. Tahap preprocessing meliputi cleaning, case folding, tokenization, dan stopword removal. Selanjutnya dilakukan pembobotan TF-IDF dan proses klasifikasi menggunakan algoritma Naïve Bayes. Pengujian dilakukan menggunakan tiga skenario pembagian data, yaitu 70:30, 80:20, dan 90:10. Evaluasi model dilakukan menggunakan Confusion matrix dengan metrik accuracy, precision, recall, dan F1-score.
Hasil penelitian menunjukkan bahwa skenario pembagian data 90:10 memberikan kinerja terbaik dengan accuracy sebesar 82,22%, precision sebesar 79,55%, recall sebesar 83,33%, dan F1-score sebesar 81,36%. Sementara itu, skenario 70:30 menghasilkan accuracy sebesar 76,67%, precision sebesar 78,90%, recall sebesar 68,25%, dan F1-score sebesar 73,17%, sedangkan skenario 80:20 memperoleh accuracy sebesar 77,78%, precision sebesar 77,56%, recall sebesar 73,81%, dan F1-score sebesar 75,61%. Hasil penelitian menunjukkan bahwa algoritma Naïve Bayes dengan pembobotan TF-IDF efektif digunakan untuk analisis sentimen komentar YouTube mengenai isu kesehatan mental.

Kata kunci: Analisis Sentimen, Kesehatan Mental, YouTube, Naïve Bayes

Mental health is an increasingly prominent issue in Indonesia. The increasing use of social media, particularly YouTube, encourages people to share their opinions and experiences regarding mental health through the comments section. These comments can be used to determine public sentiment trends using sentiment analysis techniques. This study aims to analyze the sentiment of YouTube comments on mental health issues in Indonesia using the Naïve Bayes algorithm and determine the best data sharing scenario.
The study used 900 comments from three YouTube videos on the topic of mental health obtained through a scraping process. The data was manually selected and labeled by three annotators using a majority voting method. The preprocessing stage included cleaning, case folding, tokenization, and stopword removal. Next, TF-IDF weighting and classification processes were performed using the Naïve Bayes algorithm. Testing was conducted using three data sharing scenarios: 70:30, 80:20, and 90:10. Model evaluation was conducted using a Confusion matrix with accuracy, precision, recall, and F1-score metrics.
The research results indicate that the 90:10 data split scenario yielded the best performance, with an accuracy of 82.22%, precision of 79.55%, recall of 83.33%, and an F1-score of 81.36%. Meanwhile, the 70:30 scenario resulted in an accuracy of 76.67%, precision of 78.90%, recall of 68.25%, and an F1-score of 73.17%, whereas the 80:20 scenario achieved an accuracy of 77.78%, precision of 77.56%, recall of 73.81%, and an F1-score of 75.61%. The findings demonstrate that the Naïve Bayes algorithm with TF-IDF weighting is effective for the sentiment analysis of YouTube comments regarding mental health issues.

Keywords: Sentiment Analysis, Mental Health, YouTube, Naïve Bayes