التنبؤ بالنمو الاقتصادي في سورية: دراسة مقارنة بين نماذج الانحدار الذاتي للمتوسطات المتحركة والمدعومة بالشبكات العصبية

Abstract

This study aims to enhance the accuracy of forecasting economic growth in Syria by developing a hybrid model that combines Autoregressive Integrated Moving Average (ARIMA) models and Recurrent Neural Networks (RNN). The research utilized Syrian economic growth data from the Central Bureau of Statistics for the period 1961–2022. The ARIMA model was applied using the "Auto ARIMA" function in R Studio, while the RNN model was developed using the Keras and TensorFlow libraries. The results of both models were integrated into a hybrid model (NNAR) to improve forecasting accuracy. The findings demonstrated the clear superiority of the hybrid model, achieving a Mean Error (ME) of 0.0025, a Root Mean Square Error (RMSE) of 2.886, and a Mean Absolute Percentage Error (MAPE) of 1.361%, compared to the standalone ARIMA and RNN models. The hybrid model also predicted significant fluctuations in Syria's economic growth from 2023 to 2030, with growth rates ranging between -1.247% and 3.247%. The study recommends adopting the hybrid model as a primary tool for supporting decision-makers in Syria, training technical staff in artificial intelligence techniques, and developing proactive plans to address potential economic challenges.