Prediksi Nilai Tukar IDR/USD Menggunakan Metode Hybrid LSTM-GRU

Husna A.M, Ramadhanul (2025) Prediksi Nilai Tukar IDR/USD Menggunakan Metode Hybrid LSTM-GRU. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5026211059-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2027. Download (5MB) | Request a copy

Abstract

Pasar valuta asing (forex) merupakan salah satu pasar keuangan terbesar yang memiliki volatilitas tinggi. Fluktuasi nilai tukar USD/IDR sangat berdampak pada keputusan investasi, kebijakan ekonomi di Indonesia. Prediksi nilai tukar sangat penting untuk membantu investor dalam membuat keputusan. Penelitian ini bertujuan mengembangkan model prediksi nilai tukar USD/IDR menggunakan metode hybrid LSTM-GRU. Metode hybrid LSTM-GRU memanfaatkan keunggulan Long Short-Term Memory (LSTM) dalam menangkap ketergantungan jangka panjang dan Gated Recurrent Unit (GRU) dalam efisiensi komputasi. Model dibangun dengan menggunakan data historis harian USD/IDR dan XAU/USD periode 2014-2024, serta melalui serangkaian tahap praproses, analisis seasonalitas, prediksi, dan evaluasi. Hasil eksperimen menunjukkan bahwa model hybrid LSTM-GRU terbaik dengan skenario multivariat menggunakan train test split 70%-30%, sequence length 10, batch size 32, optimizer adam, serta jumlah unit LSTM dan GRU 256 menunjukkan performa terbaik dengan nilai MAPE 0,28% dan RMSE 59,51. Model ini menunjukkan efektivitas pendekatan hybrid LSTM-GRU untuk prediksi nilai tukar USD/IDR. Penambahan variabel XAU/USD meningkatkan akurasi sekiatr 6,67%, meskipun tidak signifikan dan berdampak pada waktu eksekusi yang lebih lama. Hasil ini juga menunjukkan bahwa nilai tukar emas sebagai komoditas global memiliki korelasi dengan pergerakan USD/IDR dan sebaliknya. Temuan ini diharapkan dapat membantu para investor memahami dinamika pasar forex dan membuat keputusan investasi yang lebih baik.
===================================================================================================================================
The foreign exchange (forex) market is one of the largest financial markets with high volatility. Fluctuations in the USD/IDR exchange rate have a significant impact on investment decisions and economic policies in Indonesia. Exchange rate prediction is crucial to assist investors in making decisions. This study aims to develop a prediction model for the USD/IDR exchange rate using the hybrid LSTM-GRU method. The hybrid LSTM-GRU method leverages the advantages of Long Short-Term Memory (LSTM) in capturing long-term dependencies and Gated Recurrent Units (GRU) in computational efficiency. The model is built using historical daily data of USD/IDR and XAU/USD from 2014 to 2024, including a series of preprocessing, seasonality analysis, prediction, and evaluation steps. The experimental results show that the best hybrid LSTM-GRU model with a multivariate scenario, using a 70%-30% train-test split, sequence length of 10, batch size of 32, the Adam optimizer, and 256 units for both LSTM and GRU, achieves the best performance with a MAPE of 0.28% and RMSE of 59.51. This model demonstrates the effectiveness of the hybrid LSTM-GRU approach for predicting the USD/IDR exchange rate. The inclusion of the XAU/USD variable improved accuracy by approximately 6.67%, although it was not significant and resulted in longer execution times. These results also indicate that the gold exchange rate, as a global commodity, has a correlation with the USD/IDR exchange rate and vice versa. This finding is expected to help investors understand forex market dynamics and make better investment decisions.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Forex, Long Short-Term Memory, Gated Recurrent Unit, Prediksi, Prediction.
Subjects:	H Social Sciences > HB Economic Theory > Economic forecasting--Mathematical models. Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 57201-(S1) Undergraduate Thesis
Depositing User:	Ramadhanul Husna A.M
Date Deposited:	21 Jan 2025 07:19
Last Modified:	22 Jan 2025 05:31
URI:	http://repository.its.ac.id/id/eprint/116490

Actions (login required)

View Item