PRINCIPAL COMPONENT ANALYSIS (PCA) PARA A AVALIAÇÃO DE DADOS QUÍMICOS E GERAÇÃO DE HEAT MAPS: UM TUTORIAL

Abstract

PRINCIPAL COMPONENT ANALYSIS (PCA) FOR CHEMICAL DATA EVALUATION AND HEAT MAPS PREPARATION: A TUTORIAL. This tutorial shows a step-by-step guide on handling big datasets using principal component analysis (PCA). A dataset of chemical elements’ concentration, emission spectrum, and energy-dispersive X-ray fluorescence (EDXRF) of e-waste were used as examples. Five routines were proposed to apply data processing and PCA calculation focusing data from laser-induced breakdown spectroscopy (LIBS), EDXRF, and heat maps preparation. These routines can be used in various softwares such as MatLab, Octave, R, and Python. PCA was applied in three examples; the first was for concentrations, and the other two were for spectra. An example of heat maps assembling a hyperspectral image of a printed circuit was also described. In addition, a playlist was created on YouTube using the available examples. Therefore, with this tutorial, it may be possible to learn how to deal with a large volume of data by applying PCA. The authors hope to contribute to those researching in the area.

Keywords

• exploratory analysis
• data mining
• data visualization
• direct solid sample analysis
• laser
• X-ray fluorescence