Ashadi, Yudistira (2025) Pembuatan Model Prediksi Yield Pasar Surat Utang Negara (SUN) Menggunakan Sentimen Analisis dan LSTM. Masters thesis, Institut Teknologi Sepuluh Nopember.
![]() |
Text
6032222041-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2027. Download (3MB) | Request a copy |
Abstract
Surat Utang Negara (SUN) merupakan instrumen penting untuk membiayai defisit APBN dan mengelola portofolio utang negara. Dinamika pasar SUN dipengaruhi berbagai faktor seperti kebijakan publik, peristiwa geopolitik dan rumor pasar. Namun, model prediksi yang ada saat ini masih terbatas pada variabel ekonomi makro konvensional tanpa mempertimbangkan sentimen media. Penelitian ini menggunakan text mining dan natural language processing (NLP) untuk mengukur sentimen pasar menggunakan sumber informasi tekstual dari platform-platform berita nasional. Metodologi yang digunakan meliputi pendekatan leksikal menggunakan VADER dan pembelajaran mesin dengan Long Short-Term Memory (LSTM). Analisis dari 220.904 artikel periode 2012-2024 menunjukkan adanya hubungan asimetris yang signifikan antara sentimen dan imbal hasil, di mana pasar cenderung bereaksi lebih tidak terprediksi terhadap berita negatif dibandingkan berita positif atau netral. Model prediksi LSTM dibuat untuk memprediksi imbal hasil menggunakan data sentimen harian. Model yang dikembangkan menunjukkan performa yang baik dengan nilai Mean Squared Error (MSE) 0,002450, Mean Absolute Error (MAE) 0,0356868, Root Mean Squared Error (RMSE) 0,049500, dan Mean Absolute Percentage Error (MAPE) 0,519474. Model berhasil menangkap baik tren jangka panjang maupun fluktuasi jangka pendek dalam pergerakan imbal hasil SUN. Penelitian ini memberikan pemahaman mendalam tentang pengaruh sentimen pasar terhadap pergerakan SUN di Indonesia dan menghasilkan model prediksi yang dapat membantu investor, pelaku pasar, dan pembuat kebijakan dalam membuat keputusan investasi dengan mempertimbangkan sentimen media.
=====================================================================================================================================
Government Bond is an important instrument to finance the state budget deficit and manage the country's debt portfolio. The dynamics of the government bond market are influenced by various factors such as public policy, geopolitical events and market rumors. However, the current prediction model is still limited to conventional macroeconomic variables without considering media sentiment. This study uses text mining and natural language processing (NLP) to measure market sentiment from textual information sources of news channels. The methodology used includes lexical approaches using VADER and machine learning using Long Short-Term Memory (LSTM). Analysis of 220,904 articles from 2012-2024 shows a significant asymmetric relationship between sentiment and yields, where the market tends to react more unpredictably to negative news compared to positive or neutral news. The developed LSTM prediction model is built to predict yields using daily sentiment data. The prediction model shows good performance with Mean Squared Error (MSE) of 0.002450, Mean Absolute Error (MAE) of 0.035686, Root Mean Squared Error (RMSE) of 0.049500, and Mean Absolute Percentage Error (MAPE) of 0.519474. The model successfully captures both long-term trends and short-term fluctuations in SUN yield movements. This research provides deep insights into the influence of market sentiment on SUN movements in Indonesia and produces a prediction model that can help investors, market participants, and policymakers in making investment decisions by considering media sentiment.
Item Type: | Thesis (Masters) |
---|---|
Uncontrolled Keywords: | Surat Utang Negara; Sentiment Analysis; yield prediction. |
Subjects: | Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry) |
Divisions: | Interdisciplinary School of Management and Technology (SIMT) > 61101-Master of Technology Management (MMT) |
Depositing User: | Yudistira Ashadi |
Date Deposited: | 27 Jan 2025 01:36 |
Last Modified: | 27 Jan 2025 01:36 |
URI: | http://repository.its.ac.id/id/eprint/116919 |
Actions (login required)
![]() |
View Item |