Forecasting Algerian time series: a comparative study of ANN and SARIMA models

Accurate time series forecasting is essential for informed decision-making in economic planning, financial management, and environmental monitoring. Traditional linear models such as the Seasonal Autoregressive Integrated Moving Average (SARIMA) are widely used but often fail to capture the nonlinear and complex dynamics inherent in many real-world datasets. In recent years, Artificial Neural Networks (ANNs) have emerged as a powerful alternative, capable of modeling such complexities without relying on rigid assumptions. This study applies ANN models to three Algerian time series: Gross Domestic Product (GDP), the US Dollar Algerian Dinar (USD/DZD) exchange rate, and monthly average temperature. The forecasting performance of ANN models is benchmarked against SARIMA models. Empirical results demonstrate that ANNs consistently outperform SARIMA models in terms of predictive accuracy across all datasets, highlighting their robustness and adaptability in diverse forecasting contexts.