Semen, Geraldo Enrico (2024) INTERPRETASI HASIL PREDIKSI HASIL VOLATILITAS PASAR SAHAM LQ45 DENGAN SHAPLEY ADDITIVE EXPLANATION (SHAP) PADA MODEL HIBRID LONG SHORT TERM MEMORY - GATED RECURRENT UNIT (LSTM-GRU). Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5002201098-undergraduate_thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5002201098-undergraduate_thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (14MB) | Request a copy |
Abstract
Penelitian ini mengkaji penerapan model hibrid Long Short-Term Memory dan Gated Recurrent Unit (LSTM-GRU) yang diinterpretasikan menggunakan SHapley Additive exPlanations (SHAP) untuk memprediksi dan menganalisis volatilitas harga saham Indeks LQ45. Fokus utama penelitian adalah mengungkap perbedaan kontribusi fitur dalam prediksi harga saham (Open, High, Low, Close) selama periode volatilitas tinggi dan rendah, serta menginvestigasi bagaimana perubahan kondisi pasar mempengaruhi efektivitas indikator teknikal. Model LSTM-GRU dilatih menggunakan data historis Indeks LQ45 dari periode 1 Februari 2019 hingga 31 Januari 2024. Kinerja model dievaluasi menggunakan Mean Absolute Percentage Error (MAPE), mencapai akurasi tinggi dengan MAPE berkisar antara 0.63% hingga 0.86% untuk prediksi berbagai komponen harga. Volatilitas pasar diestimasi menggunakan metode Garman-Klass-Yang-Zhang (GKYZ). Analisis SHAP mengungkapkan bahwa pada periode volatilitas tinggi, indikator momentum seperti Relative Strength Index (RSI) memiliki pengaruh dominan terhadap prediksi harga. Sebaliknya, pada periode volatilitas rendah, harga historis terkini menjadi faktor yang lebih berpengaruh. Volume perdagangan menunjukkan pengaruh moderat namun konsisten terhadap prediksi harga dan volatilitas di kedua periode. Hasil penelitian ini memberikan wawasan berharga tentang dinamika pasar saham Indonesia, khususnya Indeks LQ45, dan mendemonstrasikan potensi integrasi teknik deep learning dengan metode interpretabilitas dalam analisis pasar keuangan. Temuan ini dapat berkontribusi pada pengembangan strategi trading yang lebih adaptif dan pemahaman yang lebih baik tentang perilaku pasar saham dalam berbagai kondisi volatilitas.
============================================================
This research examines the application of a hybrid Long Short-Term Memory and Gated Recurrent Unit (LSTM-GRU) model interpreted using SHapley Additive exPlanations (SHAP) to predict and analyze stock price volatility of the LQ45 Index. The main focus of the study is to uncover the differences in feature contributions for stock price prediction (Open, High, Low, Close) during periods of high and low volatility, and to investigate how changing market conditions affect the effectiveness of technical indicators. The LSTM-GRU model was trained using historical data of the LQ45 Index from February 1, 2019, to January 31, 2024. The model's performance was evaluated using Mean Absolute Percentage Error (MAPE), achieving high accuracy with MAPE ranging from 0.63% to 0.86% for predicting various price components. Market volatility was estimated using the Garman-Klass-Yang-Zhang (GKYZ) method. SHAP analysis revealed that during high volatility periods, momentum indicators such as the Relative Strength Index (RSI) have a dominant influence on price predictions. Conversely, during low volatility periods, recent historical prices become more influential factors. Trading volume showed a moderate but consistent influence on price and volatility predictions in both periods. The results of this study provide valuable insights into the dynamics of the Indonesian stock market, particularly the LQ45 Index, and demonstrate the potential of integrating deep learning techniques with interpretability methods in financial market analysis. These findings can contribute to the development of more adaptive trading strategies and a better understanding of stock market behavior under various volatility conditions.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | Volatilitas, Garman-Klass-Yang-Zhang, Long-Short Term Memory (LSTM), Gated Recurrent Unit (GRU), Shapley Additive exPlanation (SHAP) ============================================================ Volatility, Garman-Klass-Yang-Zhang, Long-Short Term Memory (LSTM), Gated Recurrent Unit (GRU), Shapley Additive exPlanation (SHAP) |
| Subjects: | Q Science Q Science > QA Mathematics Q Science > QA Mathematics > QA336 Artificial Intelligence Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) |
| Divisions: | Faculty of Science and Data Analytics (SCIENTICS) > Mathematics > 44201-(S1) Undergraduate Thesis |
| Depositing User: | Geraldo Enrico Semen |
| Date Deposited: | 07 Aug 2024 17:35 |
| Last Modified: | 07 Aug 2024 17:35 |
| URI: | http://repository.its.ac.id/id/eprint/113107 |
Actions (login required)
 |
View Item |