<![CDATA[Toko Oleh-Oleh Dury Weleri faces challenges in inventory management and promotional strategy due to the limitations of a conventional sales data recording system. This study aims to classify products based on sales performance using the K-Means Clustering algorithm by analyzing total sales, average sales, and remaining stock attributes. The optimal number of clusters was determined through a combination of the Elbow Method, Silhouette Score, and Davies-Bouldin Index, resulting in four main clusters. Cluster 0 consists of products with low sales and high stock (indicating potential overstock), Cluster 1 includes products with high sales but low stock (key products), Cluster 2 comprises products with moderate sales and relatively high stock (requiring light promotions), and Cluster 3 contains products with low sales and very low stock (likely seasonal or low-priority items). The clustering evaluation produced a Silhouette Score of 0.47336 and a DBI of 0.72644, indicating a reasonably good grouping quality. Interactive visualization via Streamlit provided strategic insights for decision-making regarding restocking and promotional planning. These findings are expected to support management in optimizing inventory control, improving operational efficiency, and developing more targeted sales strategies.]]>
Copyrights © 2025
