Title:
A HYBRID ADAPTIVE QUANTILE REGRESSION AND ENSEMBLE LEARNING FRAMEWORK FOR ROBUST TIME-SERIES FORECASTING

Authors:
Dr Aladdin Mahmood Karim Al-Khafaji

Abstract:
Accurate time-series forecasting is essential for decision-making in domains such as finance, energy management, healthcare, and supply chain optimization. Traditional forecasting approaches often struggle with non-stationary data, outliers, and uncertainty, leading to unreliable predictions in dynamic environments. This study proposes a robust forecasting framework that combines Adaptive Quantile Regression (AQR) with Ensemble Learning techniques to improve predictive accuracy and uncertainty estimation. Adaptive Quantile Regression enables the estimation of conditional quantiles while dynamically adjusting to changes in data distributions. Ensemble models leverage the strengths of multiple forecasting algorithms to reduce variance and improve generalization. The proposed framework generates probabilistic forecasts that provide both point estimates and prediction intervals, allowing decision-makers to assess future risks more effectively. Experimental evaluation using benchmark time-series datasets demonstrates that the hybrid approach outperforms conventional statistical and machine-learning methods in terms of forecasting accuracy, robustness, and interval coverage. The findings suggest that integrating adaptive probabilistic modelling with ensemble learning provides a powerful solution for forecasting in complex and evolving environments.

Keywords:
Time-Series Forecasting, Quantile Regression, Ensemble Learning, Probabilistic Forecasting, Machine Learning, Adaptive Models, Forecast Uncertainty

DOI:
https://doi.org/10.37500/IJESSR.2026.9322

Date of Publication: 25-06-2026