ANALYSI MI STASIMON CHRONOSEIRON KAI EFARMOGES TOUS

In the present study, the stochastic time series models are examined and the ARIMA model in forecasting GDP per quarter as well as the seasonal SARIMA model, which proves to be more appropriate of the two models. First, the basic statistical characteristics of the time series are described and then the ARIMA and SARIMA models are analyzed. The Box Jenkin method is used for the optimal selection of the parameters p, d, q in the ARIMA model. The evaluation of the forecast models is performed with both Akaike Information Criteria (AIC) and Bayesian Information Criteria (BIC) indices for the ARIMA models and with the Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) as well as with the evaluation of the Auto Correlation Function (ACF) and Partial ACF (PACF) plots. Finally, the procedure followed for the construction of the ARIMA, SARIMA models, for the execution of the forecasts on the selected GDP data per quarter is analyzed step by step