Peramalan Data Runtun Waktu Menggunakan Empirical Mode Decomposition (EMD) DAN Long Short-Term Memory (LSTM) (Studi Kasus: Harga Minyak Mentah WTI)
NOVIA DYAH AYU TRIASWARI, Dr. Noorma Yulia Megawati, S.Si., M.Sc., M.Act.Sc.
2025 | Skripsi | STATISTIKA
Time series forecasting plays a crucial role in various sectors, including finance, economics, energy, and risk management. One of its applications is predicting crude oil prices, which have a significant impact on both the global and national economies. Fluctuations in global oil prices affect national revenue and energy subsidy burdens, necessitating accurate forecasting models to assist governments in making strategic policy decisions. This study proposes a combination of the Empirical Mode Decomposition (EMD) and Long Short-Term Memory (LSTM) methods to improve the accuracy of crude oil price forecasting. EMD is used to decompose the data into several Intrinsic Mode Functions (IMFs) and residuals, enabling the capture of complex patterns in time series data. LSTM, as one of the deep learning algorithms adept at capturing long-term patterns, is employed to learn the patterns of each IMF. The LSTM model construction involves testing combinations of the lookback window size and the number of neurons in the hidden layer. The results indicate that the hybrid EMD-LSTM model outperforms the standard LSTM model, achieving an RMSE of 0.9797, MAE of 0.7432, and MAPE of 0.9030%.
Kata Kunci : Peramalan Data Runtun Waktu, Harga Minyak Mentah, Empirical Mode Decomposition, Long Short-Term Memory