Forecasting the Inflation Rate in Syria for the Period 1961–2030 Using the Hybrid CNN–LSTM Model

Abstract

This study explores the evolution of the inflation rate in Syria and forecasts its future trends through 2030 using the hybrid CNN–LSTM model. It relies on annual data for the period 1961–2023 within an economic context characterized by volatility, nonlinearity, and recurrent shocks. The statistical analysis shows that the time series does not follow a normal distribution and is characterized by positive skewness and excess kurtosis, indicating the presence of extreme values and heavy tails, which justifies the use of nonlinear modeling. The data were normalized and reformulated using a rolling time window, then divided into a training sample and a testing sample at ratios of 80% and 20%, respectively. The results show that the hybrid model achieved better forecasting performance than the ARIMA model, with good generalization ability. The future forecasts also indicate that inflation is likely to remain relatively high during the period 2024–2030, within prediction intervals that reflect a degree of uncertainty. The study concludes that the proposed model provides an appropriate framework for forecasting inflation in unstable economic environments and supports the construction of more realistic scenarios for economic decision-making