Innovative machine learning approaches for complexity in economic forecasting and SME growth : a comprehensive review

Economic forecasting and small and medium-sized enterprises (SMEs) growth prediction have become essential tools for guiding policy, business strategy, and economic development in an increasingly data-driven world. This paper reviews recent advancements in economic regression and SME growth forecasts, with a focus on the application of machine learning (ML) techniques. Specifically, the findings highlight that the integration of ensemble methods and deep learning models has achieved significant improvements in prediction accuracy, while interpretability tools such as SHAP and LIME enhance transparency and user trust. It provides a structured analysis of diverse methodologies that includes ensemble methods, deep learning models, and interpretability tools to evaluate their effectiveness and limitations in addressing the complexities of economic and SME data. This review categorizes studies by regional focus to highlight unique challenges in different economic landscapes and the adaptability of various forecasting models. Key challenges-such as imbalanced data, feature selection, and the integration of real-time data-were identified as critical factors for enhancing prediction reliability and applicability. By comparing existing surveys and identifying gaps, this review presents actionable insights and proposes future research directions that emphasize the need for integrative models that combine Explainable Artificial Intelligence (XAI) with cross-regional data fusion for more accurate and adaptable economic forecasts. These integrative models have the potential to achieve greater regional generalizability by the offering of better decision-making tools for policymakers. The findings underscore the transformative role of ML and XAI in economic forecasting and offer valuable guidance for researchers and decision-makers to optimize forecasting models for business growth and economic planning.