Prasetyo, Yoga Dwi (2026) Implementasi Model Explainable Boosting Machine dengan Pendekatan Walk-Forward Validation untuk Prediksi Harga Emas di Indonesia. Undergraduate thesis, UPN Veteran Jawa Timur.
![[img]](https://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text (Cover)
22083010055.-cover.pdf Download (2MB) |
|
Text (BAB I)
22083010055.-bab1.pdf Download (250kB) |
|
Text (BAB II)
22083010055.-bab2.pdf Restricted to Repository staff only until 2029. Download (649kB) |
|
Text (BAB III)
22083010055.-bab3.pdf Restricted to Repository staff only until 2029. Download (858kB) |
|
Text (BAB IV)
22083010055.-bab4.pdf Restricted to Repository staff only until 2029. Download (6MB) |
|
Text (BAB V)
22083010055.-bab5.pdf Download (196kB) |
|
Text (Daftar Pustaka)
22083010055.-daftarpustaka.pdf Download (187kB) |
|
Text (Lampiran)
22083010055.-lampiran.pdf Restricted to Repository staff only until 2029. Download (344kB) |
Abstract
Gold prices are one of the investment instruments widely used as a safe-haven asset. Gold price movements are influenced by various macroeconomic factors, such as the USD/IDR exchange rate, the Jakarta Composite Index (IHSG), and the price of Brent crude oil. This study aims to develop an accurate and interpretable gold price prediction model using the Explainable Boosting Machine (EBM). The predictor variables used include USD/IDR, IHSG, Brent, and temporal features derived from feature engineering, while the gold price is used as the target variable. Model evaluation was conducted using Walk-Forward Validation and compared with XGBoost; the model was then tested using Pseudo Out-of-Sample Horizon Evaluation and applied to recursive forecasting over a 7-day period. The results show that the EBM model outperforms XGBoost, with an RMSE of 39.689,5348, an MAE of 25.675,1540, and a MAPE of 1.0306%. Additionally, the out-of-sample evaluation indicates a MAPE below 1% up to a 7-day forecast horizon. The results of the study indicate that EBM is capable of providing accurate and transparent gold price predictions through the Shape Function and Pairwise Interaction.
| Item Type: | Thesis (Undergraduate) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||||||
| Subjects: | Q Science > QA Mathematics | ||||||||||||
| Divisions: | Faculty of Computer Science > Departemen of Data Science | ||||||||||||
| Depositing User: | Yoga Dwi Prasetyo | ||||||||||||
| Date Deposited: | 21 Jul 2026 01:56 | ||||||||||||
| Last Modified: | 21 Jul 2026 01:56 | ||||||||||||
| URI: | https://repository.upnjatim.ac.id/id/eprint/56431 |
Actions (login required)
 |
View Item |