SONATA, MUHAMMAD RAKAN (2025) ANALISIS SENTIMEN PENGGUNA TWITTER UNTUK MENILAI OPINI PUBLIK TERHADAP PERUSAHAAN PT XYZ DENGAN METODE ALGORITMA SVM DAN ALGORITMA NAIVE BAIYES. S1 thesis, Universitas Mercu Buana Jakarta.
|
Text (HAL COVER)
01 COVER.pdf Download (591kB) | Preview |
|
![[img]](https://repository.mercubuana.ac.id/style/images/fileicons/text.png) |
Text (BAB I)
02 BAB 1.pdf Restricted to Registered users only Download (101kB) |
|
|
Text (BAB II)
03 BAB 2.pdf Restricted to Registered users only Download (201kB) |
|
|
Text (BAB III)
04 BAB 3.pdf Restricted to Registered users only Download (126kB) |
|
|
Text (BAB IV)
05 BAB 4.pdf Restricted to Registered users only Download (422kB) |
|
|
Text (BAB V)
06 BAB 5.pdf Restricted to Registered users only Download (33kB) |
|
|
Text (DAFTAR PUSTAKA)
07 DAFTAR PUSTAKA.pdf Restricted to Registered users only Download (88kB) |
|
|
Text (LAMPIRAN)
08 LAMPIRAN.pdf Restricted to Registered users only Download (243kB) |
Abstract
PT XYZ, a company in the oil and gas sector, has attracted public attention particularly regarding corporate policies and alleged violations of business ethics. This study aims to analyze public sentiment on Twitter concerning these ethics violations at Pertamina and to compare the performance of two machine learning classifiers, Support Vector Machine (SVM) and Naïve Bayes. Data were collected via web scraping from Twitter over a predefined period, then preprocessed by performing text cleaning, slang normalization, tokenization, stopword removal, and stemming using NLTK and Sastrawi. Text features were represented using TF– IDF with the top 5,000 features. Two experiments were carried out on datasets of 770 and 1,161 tweets with an 80:20 train‑test split. Model evaluation was based on accuracy, precision, recall, and F1‑score metrics. Results indicate that SVM outperforms Naïve Bayes: on the first dataset, SVM achieved 0.76 accuracy versus 0.71 for Naïve Bayes; after data expansion, SVM’s accuracy rose to 0.82 while Naïve Bayes reached 0.78. SVM’s macro‑F1 reached 0.82 compared to 0.79 for Naïve Bayes. The greatest advantage of SVM was observed in detecting positive sentiment (F1‑score of 0.90 vs. 0.87) and negative sentiment (0.89 vs. 0.87). An in‑depth analysis of negative tweets revealed issues of corruption, collusion, and lack of transparency, accounting for 36.9 % of mentions. Based on these findings, SVM is recommended as the primary method for a social media‑based public opinion monitoring system for Pertamina. Its implementation allows early detection of spikes in negative sentiment, enabling management to undertake more targeted policy interventions and communication strategies. Keywords: Sentiment Analysis, Twitter, Machine Learning, Support Vector Machine, Naïve Bayes. PT XYZ sebagai perusahaan di sektor minyak dan gas menjadi sorotan publik terutama terkait kebijakan dan pelanggaran etika bisnis yang terjadi. Penelitian ini bertujuan menganalisis sentimen publik di Twitter mengenai pelanggaran etika bisnis Pertamina serta membandingkan performa dua algoritma klasifikasi yaitu Support Vector Machine (SVM) dan Naïve Bayes. Data dikumpulkan melalui web scraping Twitter dalam rentang waktu tertentu, kemudian dipra‑proses dengan pembersihan teks, normalisasi slang, tokenisasi, penghilangan stopword, dan stemming menggunakan NLTK dan Sastrawi. Representasi fitur teks dibuat dengan TF– IDF sebanyak 5.000 fitur teratas. Dua eksperimen dilakukan pada dataset berukuran 770 dan 1.161 tweet dengan pembagian data latih‑uji skema 80 : 20. Evaluasi model mengacu pada metrik akurasi, precision, recall, dan F1‑score. Hasil menunjukkan SVM unggul dibandingkan Naïve Bayes. Pada dataset pertama akurasi SVM mencapai 0,76 dan Naïve Bayes 0,71. Setelah penambahan data akurasi SVM meningkat menjadi 0,82 sedangkan Naïve Bayes menjadi 0,78. Macro‑F1 SVM mencapai 0,82 dan Naïve Bayes 0,79. Keunggulan SVM paling menonjol pada deteksi sentimen positif (F1‑score 0,90 vs 0,87) dan sentimen negatif (0,89 vs 0,87). Analisis mendalam terhadap cuitan negatif mengungkapkan isu korupsi, kolusi, dan kurangnya transparansi dengan proporsi 36,9 %. Berdasarkan pembahasan dan kesimpulan, SVM direkomendasikan sebagai metode utama untuk sistem pemantau opini publik Pertamina di media sosial. Implementasi model ini memungkinkan deteksi dini lonjakan sentimen negatif sehingga manajemen dapat mengambil intervensi kebijakan dan strategi komunikasi yang lebih tepat sasaran. Kata Kunci: Analisis Sentimen, Twitter, Machine Learning, Support Vector Machine, Naïve Bayes
Actions (login required)
 |
View Item |
