Ilham, Farah An Naafi' (2025) Peramalan Harga Minyak Mentah Dunia Menggunakan Metode Bayesian Structural Time Series Dengan Pendekatan Hamiltonian Monte Carlo. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5006211070_Buku TA.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5006211070_Buku TA.pdf - Accepted Version Restricted to Repository staff only Download (6MB) | Request a copy |
Abstract
Harga minyak mentah merupakan indikator ekonomi global yang sangat fluktuatif dan dipengaruhi oleh berbagai faktor eksternal seperti pandemi, geopolitik, dan dinamika permintaan energi. Oleh karena itu, diperlukan metode peramalan yang mampu menangkap pola tren, musiman, serta ketidakpastian dalam data. Penelitian ini bertujuan untuk meramalkan harga minyak mentah West Texas Intermediate (WTI) menggunakan metode Bayesian Structural Time Series (BSTS) dengan estimasi parameter menggunakan pendekatan Hamiltonian Monte Carlo (HMC). Data yang digunakan merupakan harga harian minyak WTI periode 1 Februari 2020 hingga 1 Februari 2025 sebanyak 1313 observasi. Data menunjukkan karakteristik volatilitas tinggi, dengan harga minimum sebesar $11.57 dan maksimum $119.78. Distribusi data condong ke kiri dengan nilai median lebih tinggi dari rata-rata, serta memiliki rentang harga yang luas. Model BSTS dibangun menggunakan komponen tren, musiman tahunan, dan residual autoregressive. Estimasi parameter dilakukan melalui HMC dengan dua skenario, yaitu 1 rantai dan 4 rantai. Evaluasi konvergensi dilakukan melalui trace plot, histogram hasil sampling, dan Gelman-Rubin Statistic. Setelah parameter terestimasi, dilakukan peramalan dengan membentuk distribusi probabilistik. Titik utama peramalan kemudian dievaluasi akurasinya menggunakan metrik Mean Absolute Percentage Error (MAPE). Hasil penelitian menunjukkan bahwa model dengan satu rantai memberikan hasil terbaik dengan nilai MAPE sebesar 4.39% pada data testing. Penelitian ini menunjukkan bahwa kombinasi BSTS dan HMC mampu menghasilkan model peramalan yang akurat. Saran untuk penelitian selanjutnya adalah melakukan penelitian lebih lanjut mengenai Gelman-Rubinn Statistic, mengeksplorasi komponen tambahan pada BSTS serta meningkatkan jumlah iterasi dan rantai untuk memperoleh hasil yang lebih stabil.
====================================================================================================================================================================================
Crude oil prices are a highly volatile global economic indicator influenced by various external factors such as pandemics, geopolitics, and energy demand dynamics. Therefore, a forecasting method capable of capturing trend patterns, seasonality, and uncertainty in the data is necessary. This study aims to forecast West Texas Intermediate (WTI) crude oil prices using the Bayesian Structural Time Series (BSTS) method with parameter estimation through the Hamiltonian Monte Carlo (HMC) approach. The data used consists of daily WTI oil prices from February 1, 2020, to February 1, 2025, totaling 1,313 observations. The data exhibits high volatility, with a minimum price of $11.57 and a maximum of $119.78. The data distribution is left-skewed, with the median value higher than the mean, and shows a wide price range. The BSTS model is constructed using trend, annual seasonal, and autoregressive residual components. Parameter estimation is performed via HMC with two scenarios: one chain and four chains. Convergence evaluation is carried out using trace plots, histograms, and the Gelman-Rubin statistic. After parameter estimation, forecasting is done by forming probabilistic distributions. The main forecast points are determined by taking the mean and then evaluated for accuracy using the Mean Absolute Percentage Error (MAPE) metric. The results show that the model with four chains provides the best accuration, achieving a MAPE of 4.39% on the testing data. This study demonstrates that the combination of BSTS and HMC can produce accurate forecasting models. Suggestions for future research include conduct further research on the Gelman-Rubin Statistics, exploring additional components in BSTS and increasing the number of iterations and chains to obtain more stable results.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | Bayesian Structural Time Series, Hamiltonian Monte Carlo, Harga Minyak, West Texas Intermediate, Crude Oil Prices |
| Subjects: | Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry) Q Science > QA Mathematics > QA279.5 Bayesian statistical decision theory. |
| Divisions: | Faculty of Science and Data Analytics (SCIENTICS) > Actuaria > 94203-(S1) Undergraduate Thesis |
| Depositing User: | Farah An - Naafi' Ilham |
| Date Deposited: | 01 Aug 2025 08:07 |
| Last Modified: | 01 Aug 2025 08:07 |
| URI: | http://repository.its.ac.id/id/eprint/125522 |
Actions (login required)
 |
View Item |