Analisis sentimen Youtube terhadap komentar toxic pada music video "HOZIER take me to church" dengan metode naive bayes

Calvin, Calvin (2024) Analisis sentimen Youtube terhadap komentar toxic pada music video "HOZIER take me to church" dengan metode naive bayes. Bachelor thesis, Universitas Pelita Harapan.

Preview	Text (Title) Title.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (29kB) | Preview
Preview	Text (Abstract) Abstacts.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (140kB) | Preview
Preview	Text (ToC) ToC.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (533kB) | Preview
Preview	Text (Chapter1) Chapter1.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (693kB) | Preview
	Text (Chapter2) Chapter2.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (1MB)
	Text (Chapter3) Chapter3.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (3MB)
	Text (Chapter4) Chapter4.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (1MB)
	Text (Chapter5) Chapter5.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (305kB)
Preview	Text (Bibliography) Bibliography.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (628kB) | Preview
	Text (Appendices) Appendices.pdf Restricted to Repository staff only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (1MB)

Abstract

Social media platforms like YouTube have significantly evolved in this digital era. People from all over the world use YouTube to watch videos and share their opinions, comments, and feelings about the videos they watch. This research focuses on sentiment analysis to understand the patterns of negative comments that may appear in the comments section of the videos. The data to be analyzed is obtained from the YouTube API and will use the Naive Bayes algorithm, along with performance evaluation of accuracy, precision, recall, and F1-score. The classification stages in this research include text preprocessing, data training, data testing, and evaluation. The accuracy performance results obtained from the Naive Bayes algorithm using the confusion matrix were 74%. The negative precision result is 96%, neutral precision is 97%, and positive precision is 64%. The negative recall result is 57%, neutral recall is 49%, and positive recall is 99%. The negative F1-score result is 71%, neutral F1-score is 65%, and positive F1-score is 78%. These numbers show that the model is very good at identifying positive samples (99% recall), although it has challenges in detecting negative and neutral samples. High precision for the negative and neutral classes shows good prediction accuracy in these classes. Overall, the higher F1-score performance in the positive class confirms the efficiency of the model in handling predictions for the positive class. / Platform media sosial seperti Youtube sangat berkembang di era digital ini. Banyak orang dari seluruh dunia menggunakan youtube untuk menonton video dan juga berbagi pendapat, komentar, dan perasaan mereka tentang video yang di tonton. Penelitian ini berfokus pada analisis sentimen untuk memahami pola komentar negatif yang mungkin muncul di kolom komentar video tersebut. Data yang akan di analisis di dapatkan dari Youtube API dan akan menggunakan algoritma Naive Bayes dan juga akan melakukan evaluasi performa akurasi, presisi, recall, dan f1-score. Tahap-tahap klasifikasi dalam penelitian ini mencakup text-preprocessing, data training, data testing hingga evaluasi. Hasil performa akurasi yang di dapatkan dari algoritma naive bayes dengan menggunakan confusion matrix adalah 74%. Hasil presisi negatif adalah 96%, presisi netral adalah 97% dan presisi positif adalah 64%. Hasil recall negatif adalah 57%, recall netral adalah 49% dan recall positif adalah 99%. Hasil f1- score negatif adalah 71%, f1-score netral adalah 65% dan f1-score positif adalah 78%. Angka-angka ini menunjukkan bahwa model sangat baik dalam mengidentifikasi sampel positif (recall 99%), meskipun memiliki tantangan dalam mendeteksi sampel negatif dan netral. Presisi yang tinggi untuk kelas negatif dan netral menunjukkan akurasi prediksi yang baik pada kelas tersebut. Secara keseluruhan, performa F1-score yang lebih tinggi pada kelas positif menegaskan efisiensi model dalam menangani prediksi untuk kelas positif.

Actions (login required)

View Item