<![CDATA[Inventory management in the food and beverage business requires a measurable approach to reduce the risk of stock shortages and excess inventory. Selaras Coffee & Space has operational kitchen raw material data that can be utilized to identify usage patterns more objectively. This study aims to group raw materials based on usage levels by comparing K-Means, Hierarchical Clustering, and K-Medoids algorithms. The data were obtained from kitchen raw material stock opname and purchase order records for February 2026, using stock_fisik, min_stock, and qty_po as clustering attributes. The research stages included data collection, preprocessing, unique item aggregation, Min-Max normalization, clustering algorithm implementation, evaluation using Sum of Squared Errors (SSE) and Silhouette Score, and implementation of the results into a web-based system. The initial dataset consisted of 3,080 rows and was aggregated into 110 unique items. The evaluation results showed that K-Means and Hierarchical Clustering achieved an SSE value of 4.630818 and a Silhouette Score of 0.781801, indicating a strong cluster structure. K-Medoids obtained an SSE value of 11.022485 and a Silhouette Score of 0.470763. K-Means was selected as the best algorithm because it achieved optimal evaluation performance and is simpler to implement in the system. The clustering results showed that 6 items were categorized as High Usage, 7 items as Medium Usage, and 97 items as Low Usage. The results can assist management in understanding raw material usage levels as a basis for more effective inventory control. ]]>
Copyrights © 2026
