Implementasi Hybrid ARIMAX-GRU Untuk Prediksi Harga Emas Antam di Indonesia

Putri, Armalia Kusuma (2026) Implementasi Hybrid ARIMAX-GRU Untuk Prediksi Harga Emas Antam di Indonesia. Undergraduate thesis, UPN Veteran Jawa Timur.

	Text (Cover) 22083010004-cover.pdf Download (1MB)
	Text (Bab 1) 22083010004-bab1.pdf Download (176kB)
	Text (Bab 2) 22083010004-bab2.pdf Restricted to Repository staff only until 16 September 2028. Download (534kB)
	Text (Bab 3) 22083010004-bab3.pdf Restricted to Repository staff only until 16 September 2028. Download (395kB)
	Text (Bab 4) 22083010004-bab4.pdf Restricted to Repository staff only until 16 September 2028. Download (1MB)
	Text (Bab 5) 22083010004-bab5.pdf Download (156kB)
	Text (Daftar Pustaka) 22083010004-daftarpustaka.pdf Download (151kB)
	Text (Lampiran) 22083010004-lampiran.pdf Restricted to Repository staff only until 16 September 2028. Download (4MB)

Abstract

Gold prices are an investment instrument of significant value in the face of economic uncertainty, but their movements tend to be volatile and influenced by various economic factors. These conditions highlight the importance of developing forecasting methods that can better model price change patterns as a basis for providing gold price forecasts. This study aims to implement a Hybrid ARIMAX-GRU model to forecast Antam gold prices in Indonesia and compare its performance with the ARIMAX model as a baseline. The data used include Antam gold prices, exchange rates, and interest rates for the period April 2016–December 2025, comprising 3,562 observations. The research methods include data preprocessing, stationarity testing using the Augmented Dickey-Fuller (ADF) test, ARIMAX modeling, residual testing, sequence formation, residual modeling using a Gated Recurrent Unit (GRU), and combining the ARIMAX forecast results with the GRU residual forecast. The hybrid approach combines ARIMAX to model linear patterns and exogenous variables with GRU to model patterns still present in the residuals. Evaluation was conducted using RMSE, MAE, and MAPE with a one-step-ahead rolling scheme. Test results show that ARIMAX yields an RMSE of Rp43,275.73, an MAE of Rp16,195.22, and a MAPE of 0.751%, while the Hybrid ARIMAX GRU yields an RMSE of Rp40,449.14, an MAE of Rp15,213.46, and a MAPE of 0.721%. These results indicate that the Hybrid ARIMAX-GRU model has a lower prediction error rate than ARIMAX. The model was then implemented in a GUI system to support interactive prediction, visualization, and historical evaluation processes.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Trimono, Trimono	NIDN0008099501	trimono.stat@upnjatim.ac.id
Thesis advisor	Nasrudin, Muhammad	NUPTK4241774675130323	nasrudin.fasilkom@upnjatim.ac.id

Subjects:

Q Science > QA Mathematics
Q Science > QA Mathematics > QA75 Electronic computers. Computer science

Divisions:

Faculty of Computer Science > Departemen of Data Science

Depositing User:

Armalia Kusuma Putri

Date Deposited:

16 Sep 2026 06:32

Last Modified:

16 Sep 2026 09:23

URI:

https://repository.upnjatim.ac.id/id/eprint/60269