Identifikasi Pola Pergerakan Harga Beras Melalui Dekomposisi Deret Waktu Secara Ensemble

Abstract

The price of rice is one of the important indicator to determine the availability of rice. Ensemble Empirical Mode Decomposition (Ensemble EMD) is an alternative approach to analyze rice price fluctuation behavior through the process decomposition into several intrinsic mode functions (IMFs) and the residual. The EMD could be applied to a dataset eventhough that the data are nonlinear and nonstationary so it can work with the characteristics of unstable rice prices between seasons and years. The concept of ensemble is needed to optimalize the characteristic scale of the IMF by adding a series of white noise in the data. This research was carried out on the identification of the time series pattern of monthly and weekly rice price in Jakarta using Ensemble EMD. Our Ensemble EMD approach resulted three IMFs which have large contribution to volatility of monthly rice price. Those are IMFs whose mean periods of 1.1, 2.8 and 5.6 years respectively. Meanwhile, from the weekly dataset, Ensemble EMD revealed three high contribution IMFs, whose mean period of 0.5, 1.0, and 3.4 years. Trend component contributes dominantly to the price level change. It could be seen by its contribution as long as 92.40% for monthly data and 96.34% for weekly data. Furthermore, the reconstruction of fine-to-coarse for weekly price data shows that all IMFs produced by decomposition weekly rice price are included in the high frequency component. It indicates that the imbalance of supply and demand market and the weather factor still affect the stability of rice prices in Jakarta.