Ibtesam Jasem (1)* and Mohamad Anan (2)
(1) General Commission for Scientific Agricultural Research (GCSAR), Damascus, Syria.
(2) Department of Mathematical Statistics, faculty of science, Aleppo University, Aleppo, Syria.
(*Corresponding author: Dr. Ibtesam Jasem E-mail: e_sam_0 @hotmail.com).
Received: 16/05/2020 Accepted: 31/05/2020
Abstract
Forecasting economic variables in order to plan and formulate production policies and food security is one of the most important objectives of quantitative economic studies. Multiple methods may be used for the purpose of obtaining economic forecasts. In this study ARIMA models were used that combine the method of Autoregressive and Integrated Moving Averages of time series with PCA technique in order to forecast the production of citrus fruits and its plantation area in Syria. We opted for the ARIMA models as they are known for a high accuracy regarding analysis of time series. In this article, annual data for production and plantation area in Syria for the period) 1970-2018 (is being used. Further to statistical tests, ARIMA (5,2,0) turned out to be the best model for forecasting citrus production until 2021. Accuracy values for production were 85.3% in case of oranges and 88.3% in case of lemons. This excelled (substantially) accuracy of conventional models we determined. The best model for forecasting the plantation area was ARIMA (1,1,0). Accuracy was 92.7% in case of orange plantation area and 94.2% in case of lemon plantation area. Here too we determined that ARIMA modeling excelled (substantially) accuracy of conventional models.
Keywords: Orange, Lemon, PCA, Forecasting, Box-Jenkines.

Full paper in Arabic: PDF