Desenvolvimento de interface gráfica baseada na análise exploratória para dados químicos multivariados

Abstract

This dissertation aimed to present a graphical user interface based on unsupervised classification methods. This tool was built with the objective of serving as a friendly, iterative, educational and scientific computational resource, with strong visual appeal and free to use. For this, the interface uses known analysis methods such as principal component analysis (PCA) and cluster analysis (CA), as well as the use of these two combined techniques. Among the clustering models, agglomerative and partitional methods are available, where the former is used to form clusters and the latter as part of a methodology for estimating the number of clusters. This analysis tool was designed with a focus on first-order multivariate chemical data; however, it can be used to treat data from different sources. For its evaluation, a set of real chemical data with three different classes was used, and the entire procedure used was presented. In addition, the interface enables data processing, showing itself as a potential independent tool for data analysis. The graphical results presented here are useful and innovative, serving as theoretical and practical support for future chemometric applications.