Accurate forecasting of Gross Domestic Product (GDP) is fundamental to economic policy-making and decision making. This study proposes a new integrated forecasting approach which combines the statistical rigor of the Autoregressive Integrated Moving Average (ARIMA) model with the flexibility of Polynomial Classifiers (PCs). This combined approach leverages the benefits of both models. ARIMA effectively captures linear time dependencies, while PCs can model nonlinear patterns with explicit functional forms. The study evaluates the proposed model on five real GDP datasets: Vietnam, China, the United Kingdom, the United States, and Cambodia. Experimental results demonstrate that the ARIMA-PC-based hybrid scheme yields more accurate results than the standalone models across four of five datasets. The accuracy is evaluated on criteria including MAE (mean absolute error), RMSE (root mean square error), and CV(RMSE) (coefficient of variation of RMSE). The results also show that the experiment on the Vietnam GDP dataset achieved MAE of 1.4471 and CV(RMSE) of 30.37%, which outperformed the two ARIMA and PC models when used separately. The only exception is that when experimenting on the Chinese dataset, PCs used independently achieved superior accuracy. The outcomes demonstrate that the hybrid technique is capable of enhancing the forecasting precision of economic time series.