A Fuzzy Inference System for Early Detection of Unproductive Stock in Retail Supply Chains

Dead stock—unsold inventory with declining market relevance—poses significant challenges for retailers, especially in sectors like consumer electronics and home appliances, where financial impact per unit is high. Existing classification methods often fail to capture early signals of declining performance, such as ageing, decreasing turnover, or prolonged inactivity. This paper proposes a fuzzy inference system (FIS) to classify inventory items along a continuum of risk: No Dead Stock, Possible Dead Stock, and Dead Stock. The system integrates multiple inventory performance indicators, including turnover ratios, ageing, coverage, and sales quota, and is grounded in domain knowledge and refined through iterative modeling. A synthetic dataset of stock keeping unit (SKU) sales patterns—designed to reflect real-world retail conditions—was used to test the system. The results demonstrate that the FIS provides interpretable outputs and can identify early signs of stock obsolescence, offering a practical decision support tool for inventory managers.