Sianipar, Yosephine Paulina (2025) Peramalan Nilai Jakarta Interbank Spot Dollar Rate (JISDOR) Dengan Metode Particle Swarm Optimization-Extreme Gradient Boosting (PSO-XGBoost). Other thesis, Institut Teknologi Sepuluh Nopember.
![]() |
Text
5006211012-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (8MB) | Request a copy |
Abstract
Pasar valuta asing (forex) merupakan pasar global yang menentukan nilai tukar mata uang suatu negara terhadap mata uang negara lain. Nilai tukar mata uang atau kurs sepenuhnya ditentukan oleh mekanisme permintaan dan penawaran pada pasar valuta asing. Selain itu dipengaruhi juga oleh kebijakan pemerintah, tingkat inflasi, dan kebijakan moneter. Banyaknya faktor yang mempengaruhi nilai tukar menyebabkan tingkat volatilitas yang tinggi. Sejak menggunakan sistem freely floating exchange rate nilai tukar rupiah mengalami fluktuasi yang cukup signifikan, khususnya kurs USD/IDR. Bank Indonesia menggunakan Jakarta Interbank Spot Dollar Rate (JISDOR) sebagai referensi nilai tukar USD/IDR. Tingginya volatilitas nilai JISDOR menuntut adanya model prediksi yang akurat untuk membantu pelaku bisnis dan pemerintah dalam menangani risiko terkait fluktuasi dari nilai tukar. Penelitian ini berfokus pada penerapan metode Extreme Gradient Boosting (XGBoost) dengan optimasi Particle Swarm Optimization (PSO) untuk memprediksi nilai JISDOR. Adapun data yang digunakan adalah nilai JISDOR yang diperoleh melalui situs Bank Indonesia dengan periode 1 Januari 2022 hingga 31 Desember 2024. Hasil analisis plot data time series menunjukkan bahwa nilai JISDOR selama periode tiga tahun terakhir mengalami tren depresiasi, yang mencerminkan pelemahan nilai Rupiah terhadap Dolar Amerika Serikat, yang dipengaruhi berbagai faktor baik internal (domestik) maupun eksternal. Melalui penelitian ini metode PSO-XGBoost terbukti memiliki kinerja yang lebih baik dalam meramalkan nilai JISDOR dibandingkan dengan metode XGBoost tanpa optimasi. Dimana model terbaik yang terpilih adalah PSO-XGBoost dengan kombinasi input lag y_(i-1) dan y_(i-2) serta parameter particle number 50 dan iteration number 50. Model ini menghasilkan nilai MAE sebesar 53,22252, RMSE sebesar 68,63537 dan MAPE sebesar 0,33580%. Hasil optimasi dengan PSO juga membuat perubahan hyperparameter, yaitu: n_estimators menjadi 135, learning_rate (eta) sebesar 0,5, max_depth sebesar 3, subsample sebesar 0,5, sedangkan colsample_bytree dan min_child_weight tetap pada nilai 1. Adapun nilai alpha, lambda, dan gamma (min_split_loss) masing-masing berubah menjadi 4,13, 0,29 dan 0,10. Dengan adanya peramalan berdasarkan model PSO-XGBoost ini, diharapkan dapat membantu investor dan pelaku pasar valas dalam pengambilan keputusan investasi, sedangkan untuk pemerintah hasil ini diharapkan dapat membantu perumusan kebijakan ekonomi yang lebih adaptif dan responsif terhadap pasar.
============================================================================================================================================
The foreign exchange market (forex) is a global market that determines the exchange rate of a country's currency against another country's currency. Exchange rates are entirely determined by the supply and demand mechanism in the foreign exchange market. Additionally, they are influenced by government policies, inflation rates, and monetary policies. The numerous factors influencing exchange rates result in high levels of volatility. Since adopting a freely floating exchange rate system, the Rupiah exchange rate has experienced significant fluctuations, particularly the USD/IDR exchange rate. Bank Indonesia uses the Jakarta Interbank Spot Dollar Rate (JISDOR) as the reference for the USD/IDR exchange rate. The high volatility of JISDOR necessitates an accurate predictive model to assist businesses and the government in managing risks associated with exchange rate fluctuations. This research focuses on the application of the Extreme Gradient Boosting (XGBoost) method with Particle Swarm Optimization (PSO) optimization to predict JISDOR values. The data used comprises JISDOR values obtained from Bank Indonesia's website for the period from January 1, 2022, to December 31, 2024. The result of time series data plot analysis shows that the value of JISDOR during the last three-year period experienced a depreciation trend, which reflects the weakening of the Rupiah against the United States Dollar, which is influenced by various internal (domestic) and external factors. Through this research, the PSO-XGBoost method is proven to have better performance in forecasting the value of JISDOR compared to the XGBoost method without optimization. Where the best selected model is PSO-XGBoost with a combination of input lag y_(i-1) and y_(i-2) and parameters particle number 50 and iteration number 50. This model produces an MAE value of 53.22252, RMSE of 68.63537 and MAPE of 0.33580%. The optimization result with PSO also causes changes in hyperparameters, namely: n_estimators becomes 135, learning_rate (eta) of 0.5, max_depth of 3, subsample of 0.5, while colsample_bytree and min_child_weight remain at a value of 1. The values of alpha, lambda, and gamma (min_split_loss) respectively change to 4.13, 0.29 and 0.10. With the forecasting based on this PSO-XGBoost model, it is expected to help investors and foreign exchange market participants in making investment decisions, while for the government this result is expected to help formulate economic policies that are more adaptive and responsive to the market.
Item Type: | Thesis (Other) |
---|---|
Uncontrolled Keywords: | JISDOR, Prediksi, PSO, XGBoost, Prediction |
Subjects: | H Social Sciences > HA Statistics > HA30.3 Time-series analysis H Social Sciences > HG Finance > HG3881 Foreign exchange. Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry) |
Divisions: | Faculty of Science and Data Analytics (SCIENTICS) > Actuaria > 94203-(S1) Undergraduate Thesis |
Depositing User: | Yosephine Paulina Sianipar |
Date Deposited: | 29 Jul 2025 02:29 |
Last Modified: | 29 Jul 2025 02:29 |
URI: | http://repository.its.ac.id/id/eprint/122423 |
Actions (login required)
![]() |
View Item |