ARTIFICIAL NEURAL NETWORKS FOR MATERIAL CLASSIFICATION IN PURCHASING PORTFOLIO MANAGEMENT

• Autor(a) principal: Santis, Rodrigo Barbosa de
• Data de Publicação: 2017
• Outros Autores: Aguiar, Eduardo Pestana de, Goliatt, Leonardo
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Revista Interdisciplinar de Pesquisa em Engenharia
• Texto Completo: https://periodicos.unb.br/index.php/ripe/article/view/21726
• topic: Strategic sourcing. Kraljic portfolio matrix. Supervised learning. Classification problem. Artificial neural network.

Strategic sourcing is an important technique within supply chain management area that allows companies to understand their purchasing portfolio and to better take advantage of their bargain power versus key suppliers, reducing supply risks to an acceptable minimum. The Kraljic portfolio matrix is an important model for evaluating materials and services categories in two main aspects: the importance of a particular purchasing and the complexity of the supply market. Although, the classification phase can present a hard task depending on the number of materials and the complexity of a specific portfolio. In this context, the present work proposes the application of machine learning classifiers ”“ artificial neural network, extreme learning machine and K-nearest neighbors ”“ for material grouping in a data set of 1,560 active items composing the electric and electronic portfolio components of a large Brazilian transportation company. Eight features are used for classifying material purchase and maintenance criticality. The artificial neural network presented an accuracy of 94.77% for purchase and 60.41% for maintenance classification.