... such as bagging, boosting, random forests and stacked regression, IEFRI achieves lower forecasting errors under rolling-origin evaluation. The iterative residual-learning mechanism is found to be effective in systematically reducing both nonlinear and linear forecast errors across successive refinement stages...