التنبؤ بالنمو الاقتصادي في سورية باستخدام الشبكات العصبية الالتفافية (CNN) للتعلم الآلي للفترة 1961–2023

Abstract

This study aims to forecast the growth rate of Syria’s Gross Domestic Product (GDP) for the period 2024–2029 using a Convolutional Neural Network (CNN) model. The model is trained on annual GDP data spanning from 1961 to 2023, obtained from the World Bank and the Syrian Central Bureau of Statistics. The time series data were transformed into a supervised learning format using a sliding window approach. The CNN model was trained on 95% of the dataset and tested on the remaining 5%. The results demonstrate high predictive accuracy, with error metrics in the test set reaching RMSE = 2.64 and MAPE = 1.09%, indicating the model’s ability to generalize and forecast accurately in an unstable economic environment. The forecasted values for the target period showed a downward trend in economic growth, accompanied by 95% confidence intervals that reflect the degree of uncertainty in the predictions. This study represents one of the first practical applications of CNN models to Syrian economic data and highlights the potential of deep learning techniques in economic forecasting within Arab contexts.