| dc.contributor.advisor | Suhaeni, Cici | |
| dc.contributor.advisor | Dito, Gerry Alfa | |
| dc.contributor.author | NURDZANAH, YULIANTI | |
| dc.date.accessioned | 2026-08-06T23:16:47Z | |
| dc.date.available | 2026-08-06T23:16:47Z | |
| dc.date.issued | 2026 | |
| dc.identifier.uri | http://repository.ipb.ac.id/handle/123456789/177487 | |
| dc.description.abstract | Pertumbuhan pasar mobil listrik di Indonesia diikuti oleh meningkatnya aktivitas diskusi pada platform YouTube. Komentar yang dihasilkan memuat informasi mengenai isu yang dibahas serta kecenderungan sentimen pengguna, namun jumlahnya yang besar dan tidak terstruktur menyulitkan analisis secara manual. Penelitian ini bertujuan untuk memanfaatkan BERTopic untuk pemodelan topik dan GPT untuk analisis sentimen. Data yang digunakan terdiri atas 28.054 komentar YouTube yang dipublikasikan selama periode 2022–2025. BERTopic mengelompokkan komentar ke dalam enam topik utama dengan nilai topic coherence sebesar 0,558. Analisis sentimen dilakukan menggunakan pendekatan zero-shot prompting dengan GPT-5.4-mini dan memperoleh balanced accuracy sebesar 0,81 serta macro F1-score sebesar 0,84. Hasil klasifikasi menunjukkan bahwa 61,5% komentar tergolong ke dalam kategori sentimen negatif. Integrasi hasil menunjukkan bahwa topik kebijakan subsidi dan transisi energi serta aspek teknis dan infrastruktur memiliki proporsi sentimen negatif tertinggi. Hasil eksplorasi temporal menunjukkan pergeseran fokus pembahasan dari kebijakan dan transisi energi pada 2022–2023 menuju aspek teknis, industri, ekonomi, dan minat pembelian pada 2024–2025. | |
| dc.description.abstract | The growth of the electric vehicle market in Indonesia has been accompanied by increasing discussion activity on YouTube. The comments generated contain information about the issues being discussed as well as trends in user sentiment. However, their sheer volume and unstructured nature make manual analysis difficult. This study aimed to utilize BERTopic for topic modeling and GPT for sentiment analysis. The data used consisted of 28,054 YouTube comments published during the 2022–2025 period. BERTopic grouped the comments into six main topics with a topic coherence value of 0.558. Sentiment analysis was performed using a zero-shot prompting approach with GPT-5.4-mini, achieving a balanced accuracy of 0.81 and a macro F1-score of 0.84. The classification results showed that 61.5% of comments were categorized as negative sentiment. The integration of both results showed that the subsidy policy and energy transition topic, as well as the technical and infrastructure topic, had the highest proportion of negative sentiment. Temporal exploration showed a shift in discussion focus from policy and energy transition in 2022–2023 toward technical aspects, industry, economy, and purchase intention in 2024–2025. | |
| dc.description.sponsorship | | |
| dc.language.iso | id | |
| dc.publisher | IPB University | id |
| dc.title | Pemanfaatan BERTopic untuk Topic Modeling dan GPT untuk Analisis Sentimen pada Komentar YouTube terkait Isu Mobil Listrik di Indonesia | id |
| dc.title.alternative | Utilization of BERTopic for Topic Modeling and GPT for Sentiment Analysis on YouTube Comments regarding Electric Vehicle Issues in Indonesia | |
| dc.type | Skripsi | |
| dc.subject.keyword | analisis sentimen | id |
| dc.subject.keyword | bertopic | id |
| dc.subject.keyword | GPT-5.4-mini | id |
| dc.subject.keyword | mobil listrik | id |
| dc.subject.keyword | YouTube | id |
| dc.subtype | Undergraduate Theses | |