Grouping Production Goods Requirements Using the K-Means Clustering Method

Abstract

The inventory management of production goods presents several challenges, including difficulties in distinguishing between necessary and unnecessary items, leading to overstocking and manual data processing issues. Additionally, the risk of data loss can impede the data processing workflow. Data testing is conducted to evaluate the accuracy of calculations and the functionality of the applied methods. The objective is to optimize production results and inventory levels in warehouses. The K-means algorithm, known for its simplicity and effectiveness, is utilized to identify clusters within the data. The first cluster (C0) has centroids at (60.33, 70.33) and includes stock data categorized as having no potential. This cluster comprises 35 records. The second cluster (C1) has centroids at (10.94, 7.11) and includes stock data categorized as available, consisting of 15 records. Testing with the RapidMiner Studio application confirms similar insights, with each cluster containing members that are divided into two clusters, each having optimal centroid values of (60.33, 70.33) for Cluster 1 (C0) and (10.94, 7.14) for Cluster 2 (C1), and a Davies-Bouldin Index evaluation score of 0.666.