Analisis Metode K-Means Clustering pada Data Penjualan Barang NDT PT. Sandya Lestari

Abstract

NDT (Non-Destructive Testing) is a material testing technique that does not damage the object being tested, with its main products including Liquid Penetrant Testing and Magnetic Particle Testing. The accumulation of unsold stock at PT. Sandya Lestari has become an important consideration in formulating marketing and sales strategies. This study aims to analyze NDT sales data using the K-Means Clustering method to group items based on their sales level. The data used consists of 28 items with the attributes of Initial Stock (SA), Sold Stock (ST), and Final Stock (SAK) over a given period. The clustering process was carried out both manually using Euclidean distance and with the help of RapidMiner Studio software, using 3 clusters. The results show that the cluster centers converged at the third iteration, with the final result being Cluster 0 (C0) containing 25 slow-moving items, Cluster 1 (C1) containing 2 moderately-selling items, and Cluster 2 (C2) containing 1 best-selling item. The cluster validity value measured using Cluster Distance Performance (Average Within Centroid Distance) was 21,073.928. The manual calculation and the software processing produced consistent results, indicating that most items (25 of 28) are classified as slow-moving, contributing to stock accumulation. These findings can serve as a basis for developing more effective marketing strategies and inventory management for the company.

Keywords: K-Means, Clustering, Sales Data, NDT Products, PT. Sandya Lestari