“…Quantitative methods employing multivariate calibration, such as those based on partial least squares regression (PLSR) [4][5][6][7]10,[28][29][30][31][32][33][34][35][36][37][38][39][40][41][42], principal component regression (PCR) [7,32,33,41,43,44], multi-linear regression (MLR) [32] or artificial neural networks (ANN) [30,45,46] have been found. PLSR applications were used to predict the main elemental composition [36,37] and characterization [38] of samples of the jewelry industry and its application to study the matrix effects in steel samples [40,41]. Ferreira et al [46] showed that the soil analyses using a portable LIBS system with a relatively low spectral resolution was not able to provide selective lines for Cu determination by univariate calibration.…”