Agricultural commodity price analysis using ensemble empirical mode decomposition

Agricultural commodity prices are inherently noisy, nonstationary and nonlinear due to seasonality of production, inelastic demand, production uncertainty, market imperfections, and series of policy regulations. Agricultural price analysis is considered as one of the challenging areas of time series analysis. Due to multifaceted nature of commodity price series, the conventional mono-scale smoothing approaches are unable to catch its nonstationary and nonlinear properties. Empirical mode decomposition (EMD) has been proposed as a new tool for timefrequency analysis method by Huang et al. (1998), which adaptively represent nonstationary signals as sums of different components. The essence of EMD is to decompose time series into a finite set of functions named intrinsic mode functions (IMFs) and residue. Each IMF and residue have its own characteristic scale, according to which it contributing towards original data series. One of the major drawbacks of the method is frequent appearance of mode mixing. Ensemble EMD (EEMD) is a substantial improvement of EMD which can better separate the scales naturally by adding white noise series to the original time series and then treating the ensemble averages as the true intrinsic modes. In this paper, daily price data of potato in Bangalore and Delhi markets are decomposed into several independent intrinsic modes with different frequencies, indicating some interesting features of price volatility. Here both the series are decomposed into eight IMFs and one residue. Further, decomposed IMFs and residue obtained through EEMD are grouped into high frequency, low frequency and trend components, which have similar frequency characteristics, using fine-to-coarse reconstruction algorithm. These IMF and residue can be used for prediction using any traditional or artificial intelligence technique.