Published Online:January 2025
Product Name:The IUP Journal of Applied Economics
Product Type:Article
Product Code:IJAE040125
DOI:10.71329/IUPJAE/2025.16.1.85-95
Author Name:Ramesh Murthy, Ramesh Murthy and Padmalatha N A
Availability:YES
Subject/Domain:Economics
Download Format:PDF
Pages:85-95
Accurate and adaptable forecasting is pivotal for decision-making in complex and dynamic environments. While traditional models like seasonal autoregressive integrated moving average (SARIMA) and Winter-Holt’s exponential smoothing struggle with nonlinearities in evolving datasets, machine learning methods such as artificial neural networks (ANN) impose significant computational demands. Ensemble forecasting approaches combine the strengths of statistical and machine learning models, however, challenges related to scalability, interpretability, and adaptability persist. This study introduces the iterative error-based ensemble forecasting (IEBEF) model, a novel framework designed to enhance forecasting accuracy. The model combines forecasts from SARIMA, Winter-Holt’s, ANN, and multiple linear regression (MLR), leveraging a neural network meta-model for nonlinear refinement and a regression-based iterative process for error correction. With a mean absolute percentage error (MAPE) of 2.58%, the IEBEF model significantly outperforms conventional ensemble methods, including bagging (4.65%), boosting (4.85%), and random forests (5.23%).
Forecasting plays a pivotal role in decision-making across diverse domains, including energy management, healthcare, retail, and finance. By enabling organizations to anticipate trends, optimize resource allocation, and address uncertainties, forecasting serves as a foundation for strategic planning.