“…To evaluate the performance of the sensor in discrimination of all 20 amino acids, the obtained spectral responses for 10 different concentrations of 0.5, 5, 10, 15, 20, 30, 40, 50, 75, and 100 μM were used to train 10 separate LDA models. A common challenge in training an LDA model is the number of variables that should be equal to or less than the number of samples. , Therefore, at first, the dimensionality of the training set for each case (2 SEs × 600 wavelengths × 20 samples × 3 replicates) was reduced using PCA, and the first 3 PCs representing the most significant variance of each data set were used as variables for training the LDA models. Following the training step, leave-one-out cross-validation was employed to assess the validity of the models.…”