Peramalan Harga Emas Indonesia Menggunakan Model Hybrid ARIMA-LSTM

Gani, Jordan Oktavianus (2023) Peramalan Harga Emas Indonesia Menggunakan Model Hybrid ARIMA-LSTM. Other thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 06211940000060-Undergraduate_Thesis.pdf] Text
06211940000060-Undergraduate_Thesis.pdf - Accepted Version
Restricted to Repository staff only until 1 October 2025.

Download (4MB) | Request a copy

Abstract

Pada era sekarang ini, investasi merupakan salah satu pilihan bisnis yang terus berkembang. Emas sebagai salah satu alternatif investasi memiliki sejumlah keuntungan yang signifikan. Cara berinvestasi emas yang tidak rumit serta harga emas yang stabil atau meningkat secara perlahan membuat investasi emas banyak dipandang oleh investor. Selain itu, pandemi COVID-19 di Indonesia menyebabkan kenaikan harga emas yang cukup signifikan. Harga emas berfluktuasi setiap saat baik dalam menit, jam, maupun hari sehingga menyebabkan ketidakpastian harga. Fluktuasi harga emas tertinggi terjadi saat awal pandemi COVID-19 di Indonesia. Oleh karena itu, harga emas perlu diprediksi agar dapat diketahui kapan waktu terbaik untuk membeli atau menjual emas yang telah disimpan sebagai investasi. Tujuan dari penelitian ini yaitu untuk memodelkan dan melakukan peramalan pada harga emas Indonesia. Metode peramalan yang digunakan dalam penelitian ini yaitu ARIMA, LSTM, hybrid ARIMA-LSTM, dan hybrid LSTM-ARIMA. Berdasarkan analisis yang telah dilakukan, model ARIMA yang digunakan untuk memodelkan data harga emas Indonesia yaitu ARIMA(1,1,0). Model ARIMA yang dihasilkan memiliki nilai RMSE sebesar 8.511,96 dan MAPE sebesar 0,72% pada data out-sample. Pemodelan harga emas Indonesia menggunakan model Long Short-Term Memory (LSTM), diperoleh hasil bahwa jaringan LSTM dengan 1 hidden layer dengan jumlah 3 neuron merupakan jaringan LSTM yang paling optimum dalam penelitian ini. Model LSTM yang dibangun menghasilkan nilai RMSE sebesar 8.978,53 dan MAPE sebesar 0,77% pada data out-sample. Pemodelan harga emas Indonesia menggunakan model hybrid dilakukan dengan 2 skenario, yaitu model hybrid ARIMA-LSTM dan hybrid LSTM-ARIMA. Pada model hybrid ARIMA-LSTM menghasilkan nilai RMSE sebesar 8.505,71 dan MAPE sebesar 0,72% pada data out-sample. Pada model hybrid LSTM-ARIMA menghasilkan nilai RMSE sebesar 8.753,26 dan MAPE sebesar 0,75% pada data out-sample. Model terbaik untuk meramalkan harga emas Indonesia periode harian dari Januari 2013 hingga April 2023 yaitu dengan menggunakan model hybrid ARIMA-LSTM karena memiliki nilai RMSE dan MAPE yang paling kecil dibandingkan model lainnya.
=======================================================================================================================================
In this era, investment is a business option that continues to grow. Gold as an alternative investment has a number of significant advantages. The way to invest in gold that is not complicated and the price of gold that is stable or increases slowly makes gold investment widely viewed by investors. In addition, the COVID-19 pandemic in Indonesia has caused a significant increase in gold prices. The price of gold fluctuates at any time either in minutes, hours or days, causing price uncertainty. The highest fluctuation in the price of gold occurred at the start of the COVID-19 pandemic in Indonesia. Therefore, the price of gold needs to be predicted in order to know when is the best time to buy or sell gold that has been saved as an investment. The purpose of this research is to model and to forecast the Indonesian gold price. The forecasting methods used in this study are ARIMA, LSTM, ARIMA-LSTM hybrid, and LSTM-ARIMA hybrid. Based on the analysis that has been done, the ARIMA model used to model data on Indonesian gold prices is ARIMA(1,1,0). The resulting ARIMA model has an RMSE value of 8,511.96 and a MAPE of 0,72% in the out-sample data. Modeling the Indonesian gold price using the Long Short-Term Memory (LSTM) model, results in test the LSTM network with 1 hidden layer with 3 neurons as the most optimum LSTM network in this study. The built LSTM model produces an RMSE value of 8,978.53 and a MAPE of 0,77% in the out-sample data. Modeling the Indonesian gold price using the hybrid model is carried out with 2 scenarios, namely the ARIMA-LSTM hybrid model and the LSTM ARIMA hybrid model. The ARIMA-LSTM hybrid model produces an RMSE value of 8,505.71 and a MAPE of 0,72% in the out-sample data. The LSTM-ARIMA hybrid model produces an RMSE value of 8,753.26 and a MAPE of 0,75% in the out-sample data. The best model for forecasting Indonesian gold prices for the daily period from January 2013 to April 2023 is by using the ARIMA-LSTM hybrid model because it has the smallest RMSE and MAPE values compared to other models.

Item Type: Thesis (Other)
Uncontrolled Keywords: ARIMA, Harga Emas Indonesia, Hybrid, LSTM, Peramalan, Forecasting, Hybrid, Indonesian Gold Price, LSTM
Subjects: Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry)
Q Science > QA Mathematics > QA336 Artificial Intelligence
Divisions: Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User: Jordan Oktavianus Gani
Date Deposited: 15 Aug 2023 07:19
Last Modified: 15 Aug 2023 07:19
URI: http://repository.its.ac.id/id/eprint/104711

Actions (login required)

View Item View Item