APPLICATION OF FOURIER SERIES ANALYSIS FOR CRUDE OIL PRICE PREDICTION IN INDONESIA AND GLOBALLY

ERISA , MAQFIROH (2026) APPLICATION OF FOURIER SERIES ANALYSIS FOR CRUDE OIL PRICE PREDICTION IN INDONESIA AND GLOBALLY. FAKULTAS MATEMATIKA DAN ILMU PENGETAHUAN ALAM, UNIVERSITAS LAMPUNG.

[img]
Preview
 Text
ABSTRAK.pdf
Download (98kB) | Preview
[img] Text
SKRIPSI FULL.pdf
Restricted to Repository staff only
Download (3MB) | Request a copy
[img]
Preview
 Text
SKRIPSI TANPA BAB PEMBAHASAN.pdf
Download (2MB) | Preview

Abstract

Peramalan harga minyak mentah merupakan instrumen penting dalam pengambilan keputusan ekonomi global, mengingat perannya yang vital di berbagai sektor industri dan transportasi. Penelitian ini bertujuan untuk mengidentifikasi pola musiman dan siklus, serta menerapkan integrasi metodeLocally Estimated Scatterplot Smoothing (LOESS) dan Analisis Deret Fourier (FSA) untuk memprediksi harga minyak mentah di Indonesia (ICP) dan dunia (Brent). Data yang digunakan mencakup periode bulanan dari Januari 2016 hingga Desember 2025. Metodologi penelitian melibatkan dekomposisi aditif untuk memisahkan komponen tren melalui penyisiran LOESS dan pemodelan komponen siklus menggunakan deret Fourier dengan optimasi jumlah harmonik berdasarkan kriteriaGeneralized Cross Validation(GCV). Hasil menunjukkan bahwa integrasi LOESS-Fourier mampu menangkap karakteristik data non-linier dan volatil secara efektif. Model hibrida ini menghasilkan kinerja peramalan yang sangat akurat dengan nilaiMean Absolute Percentage Error(MAPE) sebesar 7,27% pada data pelatihan dan 4,07% pada data uji untuk harga minyak global. Hal ini menegaskan bahwa pendekatan FSA yang dikombinasikan dengan LOESS memiliki presisi tinggi dalam memproyeksikan fluktuasi harga minyak mentah di masa depan. Kata-kata kunci:Brent, Harga Minyak Mentah, Analisis Deret Fourier, GCV, LOESS. Crude oil price forecasting is a crucial instrument in global economic decision-making, given its vital role in various industrial and transportation sectors. This study aims to identify seasonal, cyclical patterns, apply the integration of Locally Estimated Scatterplot Smoothing (LOESS) and Fourier Series Analysis (FSA) methods to predict crude oil prices in Indonesia (ICP) and the world (Brent). The data used covers the monthly period from January 2016 to December 2025. The research methodology involves additive decomposition to separate trend components through LOESS smoothing and modelling cycle components using Fourier series with optimization of the number of harmonics based on Generalized Cross Validation (GCV) criteria. The results show that the LOESS-Fourier integration is able to capture non-linear and volatile data characteristics effectively. This hybrid model produces highly accurate forecasting performance with a Mean Absolute Percentage Error (MAPE) value of 7.27% on training data and 4.07% on test data for global oil price. This confirms that the FSA approach combined with LOESS has high precision in projecting future crude oil price fluctuations. Keywords:Brent, Crude Oil Price, Fourier Series Analysis, GCV, LOESS.

Item Type: Other
Subjects: ?? 500 ??
?? 510 ??
Divisions: Fakultas MIPA > Prodi Matematika
Depositing User: 2605171943 Digilib
Date Deposited: 19 Jun 2026 03:38
Last Modified: 19 Jun 2026 03:38
URI: http://digilib.unila.ac.id/id/eprint/100950

Actions (login required)

View Item View Item