Predicting the Gross Domestic Product (GDP) per capita over the next decade is pivotal across various strategic areas, including government policy-making, business planning, and financial markets. In recent years, the application of deep learning (DL) architectures to Time Series (TS) data has gained significant traction in addition to statistical methods especially Auto Regressive Integrated Moving Average (ARIMA). This article contributes in univariate TS analysis, employing both traditional and DL approaches with a specific focus on the GDP of Morocco. We propose a weighed ensemble model combining ARIMA, DL methods, and their components, specifically their underlying trend and seasonality. In conclusion, the combined model: (0.1 × ARIMA + 0.1 × DL + 0.700001 × DL Trend + 0.099999 x ARIMA Seasonality) demonstrated superior performance with a mean squared error (MSE) of 0.0157, a root mean squared error (RMSE) of 0.1253, and a mean absolute error (MAE) of 0.1010. The enhanced model exhibits superior accuracy and efficiency, with lower error metrics compared to both the ARIMA (3,1,6) and GRU-based models. ARIMA achieved an MSE of 0.0217, RMSE of 0.1474, and MAE of 0.1312, whereas the GRU model achieved an MSE = 0.1062, RMSE = 0.3258, and MAE = 0.2556.