Sistema de clasificación de inventarios basado en algoritmos de Machine Learning

Abstract

Effective inventory management is essential to optimize the control, storage and distribution of products within a system. In this study, an approach based on statistical analysis and machine learning algorithms was used to determine the optimal classification of items in an automotive parts inventory system. For this purpose, a database containing the spare parts sales of an automotive company over the course of a year was examined. By applying the Kmeans, Clustering Large Applications (CLARA) and Divisive Analysis (DIANA) algorithms, an optimal classification distributed in three clusters was identified. In addition, a comparative analysis with the ABC classification was performed to define the characteristics of each cluster. The results showed that the CLARA algorithm improves inventory management, allowing to optimize storage space, increase operational efficiency, reduce costs, improve customer service and make informed decisions. It can be mentioned that, some outstanding products in the resulting clusters were 2452084002, 5810159A00 and 3910045800 from clusters 1, 2 and 3 respectively; these products are relevant due to their total sales in each cluster relating their quantity, cost and sales price. This study contributes to the field of inventory management by demonstrating how the use of machine learning algorithms through statistical analysis can optimize the classification of items in the inventory, being relevant in strategic decision making through a more accurate distribution adapted to the needs of the company.