A Nested Unsupervised Learning Model for Classification of SKU’s in a Transnational Company: A Big Data Model

This work seeks to develop a nested non-supervised model that allows a transnational soft drink company to improve its decision-making for the discontinuation of products from its portfolio with the use of unsupervised models from a database with commercial and financial information for all your product line in your most important operation. The integration of different cluster methodologies through a nested non-supervised model allowed to generate a correct identification of the products that should be refined from the catalog due to financial and operational factors. Given the magnitude of the information, a cluster was integrated into a platform for data processing as well as the generation of automatic reports that could be consulted automatically through the cloud. The products identified through the nested unsupervised model made it possible to identify products that had low demand and a low contribution to the utility of the company. Removing said products from the catalog will allow maximizing the profit of the business in addition to not incurring sunk costs related to the production and distribution of low-demand products. The platform developed will allow continuous monitoring of business performance in order to automatically identify the products likely to leave the catalog. © 2021, The Author(s), under exclusive license to Springer Nature Switzerland AG.