Effective Crude Oil Prediction Using CHS-EMD Decomposition and PS-RNN Model

There is an urgent need for the prediction of oil price; in addition, for various small and large industries, individuals, and the government, it is a blessing. Nevertheless, owing to the nonlinear inherent in crude oil price (COP), market inter-relationship in oil price time series, chaotic behavior of the COP, and inherent fractality in oil price, the prediction of COP is tedious. By employing a polyharmony spline centered recurrent neural network (PS-RNN) the work presented a new framework of COP prediction to tackle those problems. By deploying the cubic hermite spline grounded on empirical mode decomposition, which offers decomposed data, the system tackles the intrinsic characteristics of fluctuating data. The knowledgeable features are extracted using simple moving average (SMA), and exponential moving average technical indicators. By utilizing the Gaussian distribution-centric Aquila optimization that aids to endure nonlinear inherent of data within low computation time, the most related features are elected. Finally, the chosen features get trained to PS-RNN. Regarding mean absolute error, root mean squared error , mean absolute percent error (MAPE), along with Symmetric MAPE, the system obtains a low relative fault along with prevents misprediction of data; in addition, sustains higher to prevailing techniques.