Peramalan Harga Minyak Mentah Dunia Dengan Menggunakan Radial Basis Function Neural Network (RBFNN)

Elizabeth, Dewi Johanna Hendrika (2021) Peramalan Harga Minyak Mentah Dunia Dengan Menggunakan Radial Basis Function Neural Network (RBFNN). Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

Text 05211740000063-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2023. Download (3MB) | Request a copy

Abstract

Minyak mentah merupakan salah satu komoditas pemegang peranan penting dalam perekonomian dunia. Setiap negara memerlukan minyak untuk berbagai aktivitas seperti konsumsi dan produksi agar dapat menggerakkan perekonomian negara tersebut. Perubahan harga minyak mentah cenderung fluktuatif sehingga perekonomian Indonesia tidak stabil. Harga kebutuhan dalam negeri tidak dapat mudah mengikuti perubahan harga minyak mentah, sehingga pemerintah perlu melakukan revisi APBN mengikuti fluktuatifnya harga agar tetap dapat menyediakan subsidi bagi masyarakat. Ketidakstabilan harga minyak mentah mendukung penelitian untuk melakukan peramalan / forecasting terhadap harga minyak mentah dunia. Forecasting adalah suatu teknik dalam mengestimasi nilai di masa depan berdasarkan data di masa lalu maupun saat ini. Model Artificial Neural Network (ANN) adalah sistem pemrosesan informasi yang memiliki karakteristik menyerupai jaringan saraf biologis. Model ANN efektif jika digunakan dalam peramalan model non linear. Salah satu model yang tergolong dalam NN adalah Radial Basis Function Neural Network (RBFNN). RBFNN dapat digunakan untuk prediksi minyak mentah karena kemampuannya mengidentifikasi hubungan non linier kompleks yang ada dalam data deret waktu berdasarkan data historis
serta untuk memperkirakan fungsi non linier dengan tingkat akurasi yang tinggi. Peramalan menggunakan model RBFNN dilakukan terhadap data univariate, dengan variabel input tunggal, maupun multivariate dengan penambahan Kurs USD/IDR sebagai variabel input. Berdasarkan penelitian yang dilakukan, diperoleh arsitektur model RBFNN terbaik pada data multivariate yaitu dengan 3 hidden nodes dan 50 maksimum iterasi pada data dengan 5 lag-time. Sedangkan, pada data univariate yaitu dengan 6 hidden nodes dan 50 maksimum iterasi dengan data 1 lag-time. Hasil penelitian menunjukkan peramalan harga minyak mentah WTI menggunakan model RBFNN terhadap data multivariate menghasilkan nilai MAPE sebesar 2,094%, dengan arti memiliki akurasi lebih baik dibandingkan nilai MAPE 2,753% pada data univariate. Hal ini menunjukkan bahwa penambahan kurs mata uang sebagai regresor mampu meningkatkan akurasi peramalan sebesar 0,659%.
=====================================================================================================
Crude oil is one of many commodities that plays an important role in the world’s economy. Every country needs oil for various activities such as consumption and production to maintain the country's economy. Changes in crude oil prices tend to fluctuate which leads to instability in the Indonesian economy. Domestic demand prices cannot easily adapt to crude oil price changes, so the government needs to revise the APBN following these fluctuations to continue to provide subsidies. The volatility of crude oil prices supports researches in forecasting the world’s crude oil prices. Forecasting is a technique in estimating future values based on past and current data. The Artificial Neural Network (ANN) model is an information processing system that has characteristics resembling a biological neural network. The ANN model is effective if it is used in forecasting non-linear models. One example of this model is the Radial Basis Function Neural Network (RBFNN). RBFNN can be used for crude oil prediction because of its ability to identify complex non-linear relationships that exist in time series data based on historical data and to estimate non-linear functions with a high degree of accuracy. Forecasting using the RBFNN model is carried out on univariate data, with a single input variable, or multivariate with the addition of the USD/IDR exchange rate as an input variable. Based on the research conducted, the best RBFNN architecture model on multivariate data is obtained with 3 hidden nodes and 50 maximum iterations on data with 5 lag times. Meanwhile, on univariate data with 6 hidden nodes and 50 maximum iterations with 1 lag-time. The research showed that forecasting WTI crude oil prices using the RBFNN model on multivariate data resulted in a MAPE value of 2.094%, which concludes that it has better accuracy than the MAPE value of 2.753% on univariate data. This confirms that the addition of the currency exchange rate as a regressor can increase the accuracy of forecasting by 0.659%.

Item Type:	Thesis (Undergraduate)
Uncontrolled Keywords:	Peramalan, Minyak Dunia, Jaringan Syaraf Tiruan, Radial Basis Function Neural Network (RBFNN), Forecasting, Crude Oil, Artificial Neural Network
Subjects:	H Social Sciences > HB Economic Theory > Economic forecasting--Mathematical models. T Technology > T Technology (General) > T174 Technological forecasting
Divisions:	Faculty of Information and Communication Technology > Information Systems > 57201-(S1) Undergraduate Thesis
Depositing User:	Dewi Johanna Hendrika Elizabeth
Date Deposited:	22 Aug 2021 02:05
Last Modified:	22 Aug 2021 02:05
URI:	http://repository.its.ac.id/id/eprint/88344

Actions (login required)

View Item