Peramalan Harga Emas ANTAM Menggunakan Metode Hybrid Autoregressive Integrated Moving Average - Gated Recurrent Unit (ARIMA-GRU)

Erlangga, Bhimo (2025) Peramalan Harga Emas ANTAM Menggunakan Metode Hybrid Autoregressive Integrated Moving Average - Gated Recurrent Unit (ARIMA-GRU). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5006211051-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Harga emas ANTAM merupakan salah satu indikator penting dalam dunia investasi dan perekonomian Indonesia. Pergerakan harganya yang cenderung fluktuatif dan kompleks menuntut penggunaan metode peramalan yang mampu menangkap pola data baik yang bersifat linear maupun non-linear. Penelitian ini bertujuan untuk membangun model hybrid antara Autoregressive Integrated Moving Average (ARIMA) dan Gated Recurrent Unit (GRU) untuk meningkatkan akurasi peramalan harga emas ANTAM. ARIMA digunakan untuk menangkap pola linear dalam data historis harga emas, sedangkan GRU digunakan untuk memodelkan residual dari ARIMA yang mengandung komponen non-linear. Data yang digunakan adalah harga harian emas ANTAM dari 2 Januari 2022 hingga 31 Desember 2024 yang diperoleh dari situs resmi logammulia.com. Proses modeling dimulai dengan transformasi logaritma dan differencing untuk mencapai stasioneritas, dilanjutkan dengan pemilihan model ARIMA terbaik berdasarkan kriteria AIC. Residual dari model ARIMA kemudian dianalisis lebih lanjut menggunakan model GRU dengan lag yang signifikan sebagai input. Evaluasi dilakukan menggunakan nilai Mean Absolute Percentage Error (MAPE). Hasil menunjukkan bahwa model ARIMA (3,1,4) menghasilkan MAPE sebesar 0,61%, sedangkan model hybrid ARIMA-GRU mampu menurunkan MAPE menjadi 0,59%. Penelitian ini menunjukkan bahwa pendekatan hybrid ARIMA-GRU mampu meningkatkan akurasi peramalan harga emas ANTAM dan memberikan kontribusi yang berarti dalam pengembangan metode prediktif untuk data time series di sektor keuangan.
=====================================================================================================================================
ANTAM gold price is one of the key indicators in Indonesia’s investment and economic landscape. Its highly volatile and complex behavior requires forecasting methods capable of capturing both linear and non-linear data patterns. This study aims to develop a hybrid forecasting model that combines Autoregressive Integrated Moving Average (ARIMA) and Gated Recurrent Unit (GRU) to improve the accuracy of ANTAM gold price prediction. ARIMA is employed to capture linear patterns in the historical price data, while GRU is used to model the ARIMA residuals, which may contain non-linear components. The dataset consists of daily ANTAM gold prices from January 2, 2022, to December 31, 2024, obtained from the official website logammulia.com. The modeling process begins with logarithmic transformation and differencing to achieve stationarity, followed by the selection of the best ARIMA model based on the Akaike Information Criterion (AIC). The residuals from the ARIMA model are further analyzed using GRU, with significant lags identified as input features. Model performance is evaluated using the Mean Absolute Percentage Error (MAPE). The results show that the ARIMA(3,1,4) model produces a MAPE of 0.61%, while the hybrid ARIMA-GRU model reduces the MAPE to 0.59%. This study demonstrates that the ARIMA-GRU hybrid approach improves the accuracy of gold price forecasting and contributes significantly to the development of predictive methods for financial time series data.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	ARIMA, Gold Price Forecasting, GRU, Hybrid Model,ARIMA, GRU, Model Hybrid, Peramalan Harga Emas.
Subjects:	H Social Sciences > HA Statistics > HA30.3 Time-series analysis
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Actuaria > 94203-(S1) Undergraduate Thesis
Depositing User:	Bhimo Erlangga
Date Deposited:	31 Jul 2025 08:26
Last Modified:	31 Jul 2025 08:26
URI:	http://repository.its.ac.id/id/eprint/124410

Actions (login required)

View Item