Improvementin product categorization from machine learning algorithms and similarity coefficients

• Autor(a) principal: Brandão, Maicom Sergio
• Data de Publicação: 2022
• Outros Autores: Godinho-Filho, Moacir, Azzolini Junior, Walther, Battissacco, Bruna Christina, Astorino Marçola, Josadak
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Revista Produção Online
• Texto Completo: https://www.producaoonline.org.br/rpo/article/view/4483
• topic: Product categorization; Machine Learning; Decision-Tree; Neural Network; Naive Bayes; Cadastro de produtos; Aprendizado de máquina; Árvore de decisão; Redes Neurais; Naive Bayes

Product categorization is an ordinary task in every business, but it involves some pitfalls when it is made by people’s judgment only. Inconsistences in product’s attributes can lead to wrong analysis, and wrongbusiness decisions at the end. Thus,the use of machine learning techniques can contribute to improve this process. The present study evaluated the use of different machine learning algorithms and problem-solving strategies in a product categorization activity based on their descriptions taking into accounta company with high speed of creation of new products, and therefore more susceptible to errors when this task is made manuallyandproposeda new process for this activity that integrates technology asa support.A new process was proposed from the best algorithm, converting the process from manual to semiautomatic. Besides the specific benefits to the company, this study also contributes to practice in unveiling the processes of building, validating and choosing machine learning models.