A Hybrid Explainable Artificial Intelligence and Artificial Neural Network Framework for Multi-Horizon Demand Forecasting and Sustainable Inventory Optimization under Demand Uncertainty

Abstract

Accurate demand forecasting and efficient inventory management are essential for resilience and sustainability in modern supply chains. Traditional forecasting methods often fail to capture demand patterns and give limited insight into decision‑making especially when markets are dynamic and uncertain. This study introduces a Hybrid Explainable Artificial Intelligence (XAI) and Artificial Neural Network (ANN) Framework for multi‑horizon demand forecasting and sustainable inventory optimization under demand uncertainty. The framework uses an Artificial Neural Network to model demand behavior and combines it with Explainable AI tools, such as SHAP (SHapley Additive exPlanations) to explain prediction results and highlight which demand‑influencing factors are most important. The resulting multi‑horizon demand forecasts are then fed into an inventory optimization module that finds the best inventory policies to reduce holding ordering and stockout costs while keeping service levels at the desired level. The framework is tested on retail demand datasets and compared with typical statistical models and cutting‑edge machine learning methods such as ARIMA, Random Forest and Long Short‑Term Memory (LSTM) networks. Model performance is measured with forecasting metrics like RMSE, MAE, MAPE and R² well as inventory‑related indicators such as total inventory cost, service level, inventory turnover and stockout frequency. Experimental results show that the hybrid framework raises demand forecasting accuracy improves inventory decision quality and supplies interpretable advice for supply chain practitioners. By blending power, with explainable decision support this approach helps create trustworthy AI‑enabled inventory management systems that support sustainable and data‑driven supply chain operations. The results give guidance to inventory managers and policymakers who want to boost operational efficiency, cut waste and strengthen supply chain resilience when business conditions are uncertain.