Fig. 3. The PLS model developed from the constituents present in coffee: (A) the scatter plot of the observed and the predicted values of odorant property [log(1/OT)] for the final PLS model. The dashed line indicates the best fit line based on test set compounds and the solid line indicates the best fit line based on the training set compounds. (B) The PLS score plot of the training set compounds using the developed PLS model. (C) The loading plot of the model descriptors and dependent variable (log(1/OT)).