Analisis Sentimen Komentar YouTube Terhadap Unjuk Rasa 2025 Dan Hubungannya Dengan IHSG Berdasarkan Metode Naïve Bayes, Regresi Logistik, Dan SVM

Rizqiliany, Shafina Nur (2026) Analisis Sentimen Komentar YouTube Terhadap Unjuk Rasa 2025 Dan Hubungannya Dengan IHSG Berdasarkan Metode Naïve Bayes, Regresi Logistik, Dan SVM. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5003221135-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (7MB) | Request a copy

Abstract

Perkembangan media sosial menjadikan platform digital sebagai sumber penting untuk memahami opini publik terhadap berbagai peristiwa sosial dan politik. Salah satu peristiwa yang mendapat perhatian luas masyarakat adalah unjuk rasa dan kerusuhan pada Agustus–September 2025 di Indonesia. Penelitian ini bertujuan menganalisis sentimen publik terhadap peristiwa tersebut, membandingkan kinerja metode Bernoulli Naïve Bayes, Regresi Logistik Biner, dan Support Vector Machine (SVM) dalam klasifikasi sentimen, serta mengkaji hubungan antara sentimen publik dan pergerakan Indeks Harga Saham Gabungan (IHSG). Data penelitian berupa 10.000 komentar YouTube yang dikumpulkan menggunakan YouTube Data API dan data IHSG periode Agustus–September 2025 yang diperoleh dari Yahoo Finance. Tahapan analisis meliputi pra-pemrosesan teks, pelabelan sentimen menggunakan InSet Lexicon yang divalidasi secara manual, pembagian data secara stratified train-test split, ekstraksi fitur TF-IDF, penyeimbangan kelas menggunakan SMOTE pada data latih, klasifikasi sentimen, dan analisis korelasi Pearson serta Spearman. Hasil penelitian menunjukkan bahwa dari 9.543 komentar yang berhasil dilabeli, sebanyak 93,17% merupakan sentimen negatif dan 6,83% sentimen positif. Berdasarkan pendekatan pembagian data yang dilanjutkan penyeimbangan kelas menggunakan SMOTE pada data latih, model Regresi Logistik Biner memberikan performa terbaik dengan accuracy sebesar 87,90%, sensitivity sebesar 79,23%, dan ROC-AUC sebesar 92,26%. Analisis korelasi Pearson menghasilkan koefisien korelasi sebesar 0,1854 dengan p-value sebesar 0,3750, sementara sebagai analisis pelengkap, korelasi Spearman menghasilkan koefisien korelasi sebesar 0,189 dengan p-value sebesar 0,3655. Kedua hasil tersebut konsisten dan menunjukkan hubungan positif sangat lemah dan tidak signifikan secara statistik antara indeks sentimen publik dan perubahan IHSG. Penelitian ini menunjukkan bahwa Regresi Logistik Biner merupakan metode paling optimal untuk klasifikasi sentimen komentar YouTube, dan terdapat hubungan positif antara sentimen publik dan pergerakan IHSG, meskipun hubungan tersebut tergolong sangat lemah karena pergerakan IHSG juga dipengaruhi oleh berbagai faktor lain di luar sentimen publik.
=======================================================================================================================================
The rapid growth of social media has made digital platforms an important source for understanding public opinion on social and political events. One such event that attracted widespread public attention was the protests and riots in Indonesia during August–September 2025. This study analyzes public sentiment toward these events, compares the performance of Bernoulli Naïve Bayes, Binary Logistic Regression, and Support Vector Machine (SVM) in sentiment classification, and examines the relationship between public sentiment and the movement of the Indonesia Composite Stock Price Index (IHSG). The study used 10,000 YouTube comments collected through the YouTube Data API and IHSG data for August–September 2025 obtained from Yahoo Finance. The analytical stages included text preprocessing, sentiment labeling using the InSet Lexicon with manual validation, stratified train-test splitting, TF-IDF feature extraction, class balancing using SMOTE on the training data, sentiment classification, and correlation analysis using Pearson and Spearman methods. Of the 9,543 successfully labeled comments, 93.17% were classified as negative sentiment and 6.83% as positive sentiment. Based on the approach of splitting the data prior to applying SMOTE on the training set, the Binary Logistic Regression model achieved the best performance, with an accuracy of 87.90%, a sensitivity of 79.23%, and a ROC-AUC of 92.26%. Pearson correlation analysis yielded a coefficient of 0.1854 with a p-value of 0.3750, while Spearman correlation, used as a complementary analysis, yielded a coefficient of 0.189 with a p-value of 0.3655. Both results were consistent, indicating a very weak, statistically insignificant positive relationship between the public sentiment index and IHSG changes. This study concludes that Binary Logistic Regression is the most effective method for classifying YouTube comment sentiment, and that a positive but very weak relationship exists between public sentiment and IHSG movements, as IHSG changes are also influenced by various factors beyond public sentiment.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Analisis Sentimen, Bernoulli Naïve Bayes, IHSG, Regresi Logistik Biner, Support Vector Machine, Bernoulli Naïve Bayes, Binary Logistic Regression, Composite Stock Price Index (IHSG), Sentiment Analysis, Support Vector Machine
Subjects:	H Social Sciences > HA Statistics H Social Sciences > HA Statistics > HA31.3 Regression. Correlation. Logistic regression analysis. H Social Sciences > HA Statistics > HA31.7 Estimation H Social Sciences > HG Finance > HG4915 Stocks--Prices Q Science Q Science > Q Science (General) Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines.
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User:	Shafina Nur Rizqiliany
Date Deposited:	28 Jul 2026 08:25
Last Modified:	28 Jul 2026 08:25
URI:	http://repository.its.ac.id/id/eprint/138803

Actions (login required)

View Item