The SARIMA modelling approach involves the following four-step iterative process: (1) identify the SARIMA(
p,d,q)(P,D,Q)S structure, (2) estimate the unknown parameter, (3) check the estimated residuals and (4) predict future
values. Residuals should satisfy the requirements of a white noise process (uncorrelated and normally distributed around a
zero mean). Further details are given in Box and Jenkins (1976) to estimate the parameters and order of the model. Here, in
order to identify the model, a set of different parameters were iteratively used to judge whether the model is appropriate and
to select the best model by comparison of prediction results