Global Prediction System of the COVID-19 Pandemic
The EEMD method can decompose the sequence into components of different frequencies, which has a better effect on the processing of non-stationary nonlinear sequences. The ARMA method can be used to predict sequences, and its prediction depends only on the time series, and no other information is needed. Owing to the advantages of the EEMD and ARMA methods, we propose a hybrid EEMD-ARMA method to predict the residuals, thereby to modify the prediction results. The EEMD method is used to decompose the seven-point smoothed original residual sequence into several components (including the residual), and then the ARMA method is used to predict each component except the residual. The predicted values can be obtained by adding up the predicted values of the components. The procedures for the hybrid EEMD-ARMA forecasting method are as follows:
Fig.1 Flow chart of prediction of the residual
Seven-point smoothing is performed on the residual sequence of the fitting result. The smoothed sequence is decomposed by the EEMD method, with the residual difference component removed. The first-order difference is calculated compared with other components, and then the ARMA model is used to predict each component. The prediction results of the appropriate components are selected for summation as the final residual prediction result.
The hybrid EEMD-ARMA method has a better effect on the high-frequency oscillations. To improve the quality of the prediction, for the residual sequence, the several days before the peak of the number of newly added cases are selected as the starting time of the sequence.
