Implementasi IndoBERT Dalam Analisis Sentimen Pada Platform X (Twitter) Untuk Prediksi Pergerakan IHSG Menggunakan Support Vector Regression

Puspitasari, Serly Diah (2026) Implementasi IndoBERT Dalam Analisis Sentimen Pada Platform X (Twitter) Untuk Prediksi Pergerakan IHSG Menggunakan Support Vector Regression. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 2043221082-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (6MB) | Request a copy

Abstract

Indeks Harga Saham Gabungan (IHSG) merupakan barometer krusial ekonomi Indonesia yang mencerminkan kepercayaan investor serta stabilitas sistem keuangan nasional. Fenomena gejolak pasar pada awal tahun 2025, yang ditandai dengan penurunan harian signifikan dan aktivasi trading halt, menunjukkan tingginya volatilitas yang dipicu oleh faktor domestik dan tekanan global. Di era digital, sentimen media sosial, khususnya platform X (Twitter), menjadi sumber data real-time yang mampu merepresentasikan opini publik dan mempengaruhi keputusan investasi. Penelitian ini bertujuan untuk menganalisis sentimen pasar dan memprediksi pergerakan IHSG dengan mengintegrasikan data historis dan opini publik. Metode penelitian ini menggunakan model IndoBERT untuk mengklasifikasikan sentimen IHSG dari platform X. Selanjutnya, prediksi pergerakan nilai IHSG dilakukan menggunakan metode Support Vector Regression (SVR), sebuah aplikasi yang efektif dalam menangani data deret waktu kontinu dan regresi kompleks. Model SVR dioptimalkan dengan algoritma Fruit Fly Optimization Algorithm (FOA). Sentimen pasar terlebih dahulu dievaluasi menggunakan metode IndoBERT yang menunjukkan tingkat akurasi sentimen keseluruhan dì atas 80%. Setelah itu, analisis menunjukkan bahwa model SVR dengan fungsi kernel RBF memberikan kinerja terbaik empat skenario pemodelan diusulkan untuk menemukan model prediksi terbaik: (1) model tanpa sentimen, (2) model dengan sentimen tanpa lag, (3) model dengan sentimen lag 1, dan (4) model dengan sentimen lag 2. Hasil akhir menunjukkan bahwa model pada skenario (3) memiliki kesalahan prediksi terendah dengan nilai MAPE dan RMSE sebesar 1,17% dan 128,4649 dibandingkan dengan model lainnya.
=====================================================================================================================================
The Jakarta Composite Stock Price Index (JCI) is a crucial barometer of the Indonesian economy that reflects investor confidence and the stability of the national financial system. The phenomenon of market turmoil in early 2025, which is characterized by significant daily declines and activation of halt trading, indicates high volatility triggered by domestic factors and global pressures. In the digital age, social media sentiment, especially platform X (Twitter), is a source of real-time data that is able to represent public opinion and influence investment decisions. This study aims to analyze market sentiment and predict the movement of JCI by integrating historical data and public opinion. This research method uses the IndoBERT model to classify the JCI sentiment from platform X. Furthermore, the prediction of the movement of the JCI value is carried out using the Support Vector Regression (SVR) method, an application that is effective in handling continuous time series data and complex regressions. The SVR model is optimized with the Fruit Fly Optimization Algorithm (FOA). The market sentiment was first evaluated using the IndoBERT method which showed an overall sentiment accuracy rate above 80%. After that, the analysis showed that the SVR model with the RBF kernel function provided the best performance four modeling scenarios were proposed to find the best prediction model: (1) a model with no sentiment, (2) a model with no lag sentiment, (3) a model with a lag sentiment of 1, and (4) a model with a lag sentiment of 2. The final results showed that the model in scenario (3) had the lowest prediction error with MAPE and RMSE values of 1.17% and 128.4649 compared to the other models.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	IndoBERT, IHSG, Sentimen, Support Vector Regression (SVR), X IndoBERT, JCI, Sentiment, Support Vector Regression (SVR), X
Subjects:	H Social Sciences > HA Statistics > HA30.3 Time-series analysis H Social Sciences > HB Economic Theory > Economic forecasting--Mathematical models. H Social Sciences > HC Economic History and Conditions > HC108 Market surveys. Q Science > QA Mathematics > QA353.K47 Kernel functions (analysis) T Technology > T Technology (General) > T57.5 Data Processing T Technology > T Technology (General) > T57.8 Nonlinear programming. Support vector machine. Wavelets. Hidden Markov models.
Divisions:	Faculty of Vocational > 49501-Business Statistics
Depositing User:	Serly Diah Puspitasari
Date Deposited:	13 Feb 2026 06:50
Last Modified:	13 Feb 2026 06:50
URI:	http://repository.its.ac.id/id/eprint/132425

Actions (login required)

View Item