3.2.2. PLS-DA
In all cases, the PLS-DA results presented fewer misclassified samples than the respective SIMCA results. As can be also observed in Tab. 3, no errors were found for both ‘‘All parameters’’ and ‘‘Elements in tea infusions’’ data sets, whereas the other two data sets achieved correct classification at 86.7%. Low R2 values for calibration and cross-validation were obtained for all data sets, despite ‘‘Elements in tea infusions’’ presenting reasonable R2 values of 0.9131 and 0.8325 for calibration and cross-validation, respectively.