Implementasi Algoritma K-Means dan Metode Elbow untuk Clustering Data Penjualan Suku Cadang Motor Multi-Platform

Irfan Ardianto (Universitas Nahdlatul Ulama Sidoarjo)
Untung Usada (Universitas Nahdlatul Ulama Sidoarjo)

Jurnal Teknik Industri Terintegrasi (JUTIN)
Vol. 9 No. 3 (2026): July

Transaction data fragmentation across multi-platform e-commerce triggers motorcycle spare part inventory imbalances. This study aims to objectively cluster spare part products to support inventory control. Employing a quantitative approach, the research analyzed a sample of 56,514 transaction records from Cuix Motorcycle, Syafik Jaya, and MSM 17 stores on Shopee, TikTok Shop, and Lazada from April 2025 to March 2026. The methodology integrated K-Means Clustering and the Elbow Method using Python. The Elbow Method identified  as the optimal cluster count, validated by a Silhouette Score of 0.761. Segmentation categorized products into three performance tiers: Fast Moving (291 items), Medium Moving (5,516 items), and Slow Moving (775 items). Recommendations include applying Safety Stock for Fast Moving items, Reorder Point systems for Medium Moving items, and Just-In-Time or bundling strategies for Slow Moving items to optimize working capital.