PLS predictions

SEC-FTIR yields the average polymer structure as a function of molecular mass, but no information on the distribution of the chemical composition within a certain size fraction. SEC-FTIR is mainly used to provide information on MW, MWD, CCD, and functional groups for different applications and different materials, including polyolefins and polyolefin copolymers [703-705]. Quantitative methods have been developed [704]. Torabi et al. [705] have described a procedure for quantitative evaporative FUR detection for the evaluation of polymer composition across the SEC chromatogram, involving a post-SEC treatment, internal calibration and PLS prediction applied to the second derivative of the absorbance spectrum. 

Fig. 15.4. Relationship between experimentally determined and PTSA-predicted Caco-2 permeability. PLS predicted permeability from PTSAs (predicted log Caco-2) is plotted versus experimentally obtained Caco-2 data (observed log Caco-2) [1 7]. The PTSAs used for the prediction were (in order of importance) PSA, fraction of total surface area that was polar...

PLS prediction can be performed in analogy to PCR by PCR PLS pred. m (p.302) already introduced earlier. For this, we need to determine the prognostic vector, vpTog. Using the results of the PLS calibration, it can be computed as ... 

Many earlier successful PLS prediction models (which in this chapter are presented as examples from industrial production processes) signify that acoustic chemometrics has matured into a proven on-line technology in the PAT domain. It is a salient point here that all were evaluated using test set validation [2]. 

In 2000 two major petrochemical companies installed process NMR systems on the feed streams to steam crackers in their production complexes where they provided feed forward stream characterization to the Spyro reactor models used to optimize the production processes. The analysis was comprised of PLS prediction of n-paraffins, /xo-paraffins, naphthenes, and aromatics calibrated to GC analysis (PINA) with speciation of C4-C10 for each of the hydrocarbon groups. Figure 10.22 shows typical NMR spectral variability for naphtha streams. Table 10.2 shows the PLS calibration performance obtained with cross validation for... 

Table 14 Root Mean Squares Errors for the NIR-PLS Predicted Values of Different Sample Attributes for the MCC Surrogate Tablets and Roller Compacted Samples...

Very little scatter was observed in the PLS predicted values for sample LOD (Table 16) and APAP concentration (Fig. 9). 

Figure 8 PLS predicted values of LOD from the NIR data collected during realtime monitoring of roller compaction at different RH. Key Diamonds, 24% RH squares, 45% RH triangles, 65% RH. Four minutes each at 7.2, 6.0, and 5.0 rpm roll speeds, respectively.

Figure 9 PLS predicted values of APAP concentration from the NIR data collected...

There are several distinctions of the PLS-DA method versus other classification methods. First of all, the classification space is unique. It is not based on X-variables or PCs obtained from PCA analysis, but rather the latent variables obtained from PLS or PLS-2 regression. Because these compressed variables are determined using the known class membership information in the calibration data, they should be more relevant for separating the samples by their classes than the PCs obtained from PCA. Secondly, the classification rule is based on results obtained from quantitative PLS prediction. When this method is applied to an unknown sample, one obtains a predicted number for each of the Y-variables. Statistical tests, such as the /-test discussed earlier (Section 8.2.2), can then be used to determine whether these predicted numbers are sufficiently close to 1 or 0. Another advantage of the PLS-DA method is that it can, in principle, handle cases where an unknown sample belongs to more than one class, or to no class at all. 

Figure 12.17 Concentration ofNH4Cl ( ) and Si-NH2 (+) on the silica gel surface, together with the PLS predictions.

Figure 12.18 Explained variance for each chemical component, as a function of the rank used in the PLS prediction model + Si-NH2 NH4Cl.

The PLS prediction is shown in figure 12.17 (dotted lines). The Standard Error of Prediction is 0.06 mmol/g for Si-NH2 and 0.03 mmol/g for NH4C1, being fairly small, considering the uncertainties in the calibration set. 

The coating allows three independent datasets leading to the prediction of PLS prediction of Fig. 12 (to be compared with that of Fig. 10) to be obtained... 

Fig. 12. PLS prediction of glucose levels in one yeast batch fermentation by a model formed and validated on glucose levels in two other independent fermentations. This experiment uses polymer coated electrodes. The rmsep is 35%...

Figure 15.1 Figures to describe the basis of novelty and margin detection to provide a confidence measure for K-PLS predictions, (a) Novelty detection, (b) margin detection for classification, and (c) margin detection for regression.

The spectral residuals, Eq, is calculated and replaces D )< owri in Eq. (36). Calculation continues until all Wfact factors have been included in the calculation of C. Alternatively, PLS prediction can be done using a calculated regression vector. A detailed discussion on the PLS method has been previously presented in the literature. ... 

Fig. 7. A data set containing (a) the solution NMR spectra for 26 mixtures of five primary alcohols. Broad line shapes are observed in the NMR due to rapid quadrupolar relaxation, making the discrimination of the individual resonance difficult. Original PLS predictions for the primary alcohols in these complex mixtures were very poor. Through the utilization of net analyte signal (NAS) methods, the combination of propanol and butanol alcohols into a single analysis group (based on carbon chain length) greatly improved the resulting (b) PLS predictions. Figure adapted from the work of Alam and Alam. ...

The PLS prediction error on the unseen validation set using all the variables is 2.1% (A= 13 PLS factors) for comparison. Thus, the prediction error is approximately the same with an almost 50% reduction in the total number of variables. A scalogram with the selected variables is shown in Fig. 13. Note that the whole of scales 2 and 3 are selected. 

Figure 4. PLS predictions of (A) combined aristeromycin and neplanocin A concentration (6 factors, MSEP 0.088) (B) aristeromycin concentration (7 factors, MSEP 0,1391) (C) neplanocin A concentration (5 factors, MSEP 0.141) from all of the S. citricolor mutant fermentation supernatants (g. I- )- Circles represent samples in the training set used to form the PLS model and squares indicate the predicted concentrations derived from unknown spectra that were not in the training set. Aristeromycin and neplanocin A concentrations were obtained by HPLC analysis.

Miners JO, Smith PA, Sorich MJ, McKinnon RA, Mackenzie PL Predicting human drug glucuronidation parameters application of in vitro and in silica modeling approaches. Annu Rev Pharmacol Toxicol 2004 44 1-25. 

Numerous additional authors have contributed to the study of hardness measurement by NIR [106-112]. Chen et al. [107] found that artificial neural networks results were comparable to PLS predictions, and Donoso... 

Figure 3 Real observation (ordinate) vi. PLS prediction (abscissa) for different time scales.

Predictions from the various models are compared with experiment in Figs. 7 and 8. The GRNN and FNN predictions give correlation coefficients better than 0.98 while the PLS prediction had a poorer correlation coefficient of 0.82 (see Fig. 7). Figure 8 illustrates the predictive capability for the reverse problem. Neural networks and PLS both perform substantially worse but still provide... 

Fig. 7c. Comparison of prediction (forward mode) with experiment PLS predictions described in the text...

Figure 43. A, bright-field image of cattle feed contaminated with 1% poultry meal. B, NIR spectra of cattle feed and poultry meal. C, results of a PLS prediction showing the pixels assigned to cattle feed (blue) and the pixels assigned to poultry meal (red).

Table 5.4 Sample-specific confidence intervals calculated for the PLS predictions of Cu in unknown commercial lubricating oils and two certified reference materials (CS2 example), values in pg Cu 1 ...

---