(2) Identifying outliers by observing training set samples residing outside the "strip's" boundaries.
Fig. 5 Predicted versus experimental PV properties for train set samples following the removal of outliers.
We propose ET estimators of the DF and use them to construct new resampling algorithms for unbalanced ranked set samples.
Optimization of NIR models, for the two algorithms, was achieved by proper selection of spectral processing, and training set samples.
The precision of the method was estimated by analysing two test set samples for six times each.
An increase in fibre draw ratio (DR) for heat set samples produced an increase in tenacity, birefringence and X-ray intensity.
For test set samples, their known activities allow to classify them as either within or outside the model's applicability domain (i.e., either within or outside the "strip").
The similarity between the test sample and the other samples, used in training, is calculated, and k most similar set samples are determined.
Only the training set samples were classified for training the Biomarker Pattern Software 5.0.2 (Ciphergen).
The assignment probabilities estimated from one-out cross validation of 48 Moroccan and 48 Spanish training set samples were plotted in the combined chart shown in Figure 1.
