<![CDATA[The food and beverage service industry operates under high demand volatility, requiring inventory systems that are both adaptive and cost-efficient. A central challenge is maintaining product availability without excessive inventory that inflates costs. The objective of this study is to develop a data-driven restocking framework that improves cost efficiency while accounting for real operational constraints. The proposed method integrates K-Means clustering with a decision tree to generate interpretable, rule-based stock recommendations. K-Means clustering was applied as an unsupervised approach to group items into risk-based zones (Green, Yellow, Red), which were then used as labels in a supervised Decision Tree model. The model achieved 99% accuracy and an F1-score of 0.93. When applied to real industry data, it reduced Total Inventory Cost (TIC) by up to 16.9% compared with the company's MOQ-based policy while preserving stable service performance. These findings demonstrate that combining clustering and rule-based machine learning provides a practical, cost-efficient, and interpretable solution for optimizing restocking decisions in complex operational environments.]]>
Copyrights © 2025
