CARACTERIZAÇÃO DO PERFIL SENSORIAL DE CREAM CHEESE: UMA ABORDAGEM DA MINERAÇÃO DE DADOS

Abstract

Data mining is currently widely used in separation and cataloging, helping understanding consumer opinions. A group analysis using data sensory results in information that can improve searches, determine changes of products, in addition to enabling evaluations for the generation of projects or reports for different analyzes and improvements. Therefore, this work presents a study that aims to characterization of the sensory profile of cream cheese through clustering of a database contains sensory information such as creaminess, graininess, butter level, etc. The main objective of this study is to create clusters that place this data in their respective groups, so that each sample is internally similar to the others in its group and among others. At the same time there must be differences between groups (clusters). For that, uses cluster analysis, decision trees, along with clustering algorithms of the R language, generating an accuracy of 94.48% in the training set and 82.11% in the test set. The final clustering result is generated through PCA and the decision.