<![CDATA[MR. GADGET store in Bengkulu faces significant challenges in managing iPhone spare parts inventory due to a manual recording system. This study proposes a data science-based solution using the K-Medoid Clustering algorithm to group data based on characteristic similarity. Utilizing a dataset of 471 products, this study compares K-Medoid with conventional partitioning methods (like K-Means), demonstrating its superior robustness against outliers by using actual data points as cluster centers. The clustering quality is evaluated using the Silhouette Score and Davies-Bouldin Index (DBI), yielding best-performing results with a Silhouette Score of 0.681 and a DBI of 0.798. The algorithm generates three main clusters: Fast-Moving, Medium-Moving, and Slow-Moving. The system's functionality is validated through Black Box Testing. The results indicate that this approach provides more accurate procurement recommendations, optimizes inventory turnover, and reduces the risk of inventory imbalance, offering a practical data-driven framework for local retail businesses.]]>
Copyrights © 2026
