Sample selection

8 Sample and variable selection in chemometrics 8.8.1 Sample selection 

Earlier in this chapter, it was mentioned that a large amount of good calibration data is required to build good empirical models. However, it was also mentioned that these data 

There are several methods that can be used to select well-distributed calibration samples from a set of such happenstance data. One simple method, called leverage-based selection, is to run a PCA analysis on the calibration data, and select a subset of calibration samples that have extreme values of the leverage for each of the significant PCs in the model. The selected samples will be those that have extreme responses in their analytical profiles. In order to cover the sample states better, it would also be wise to add samples that have low leverage values for each of the PCs, so that the center samples with more normal analytical responses are well represented as well. Otherwise, it would be very difficult for the predictive model to characterize any non-linear response effects in the analytical data. In PAC, where spectroscopy and chromatography methods are common, it is better to assume that non-linear effects in the analytical responses could be present than to assume that they are not. 

Another useful method for sample selection is cluster analysis-based selection.3 4,67 in this method, it is typical to start with a compressed PCA representation of the calibration data. An unsupervised cluster analysis (Section 8.6.3.1) is then performed, where the algorithm is terminated after a specific number of clusters are determined. Then, a single sample is selected from each of the clusters, as its representative in the final calibration data set. This cluster-wise selection is often done on the basis of the maximum distance from the overall data mean, but it can also be done using each of the cluster means instead. 

The cluster analysis-based method of sample selection is very useful when one wants to ensure that at least one sample from each of a known number of subclasses is selected for calibration. However, one must be careful to specify a sufficient number of clusters, otherwise all of the subgroups might not be represented. It is always better to err on the side of determining too many clusters. The specification of a number of clusters that is much greater than the number of natural groups in the data should still result in well-distributed calibration samples. 

Several PAT calibration strategies, especially those that are intended to support inverse calibration methods, rely heavily on data that is routinely collected from the deployed analyzer, as opposed to data collected from carefully designed experiments. Such data, often called happenstance data , can be very inexpensive. 

---

Bromochlorofluoromethane IS a known compound and samples selectively enriched in each enantiomer have been described in the chemi cal literature In 1989 two chemists at Polytechnic Uni versity (Brooklyn New York) described a method for the preparation of BrClFCH that IS predominantly one enantiomer... 

Absolute configurations of the isoxazolidines obtained in the nitrone cydoaddition reactions described in Schemes 7.21 and 7.22 were determined to be 3S,41 ,5S structure by comparison of the optical rotations as well as retention times in a chiral HPLC analysis with those of the authentic samples. Selection of the si face at C/ position of 3-crotonoyl-2-oxazolidinone in nitrone cydoadditions was the same as that observed in the Diels-Alder reactions of cyclopentadiene with 3-croto-noyl-2-oxazolidinone in the presence of the J ,J -DBF0X/Ph-Ni(C104)2-3H20 complex (Scheme 7.7), and this indicates that the s-cis conformation of the dipolaro-phile has participated in the reaction. 

Fe4Sni2S32) to 2.319 A in the Si-doped compound studied here. This may be due to the doping of Si atoms. The size of Si" is smaller in comparison to the size of Cu and this would lead to the smaller Cu-S bond length in the Si-doped compound [22]. The Fe/Sn-S distances do not show much change, being 2.536 A in the pure compound and 2.5365 A in the Si-doped sample. Selected bond distances are given in Table 15.2. 

Guidelines for the methods of sample collection of crop samples are detailed in the Codex Alimentarius, but generally in most instances crop samples should be representative of the crop being grown. As a general rule, the quantity of sample required is a minimum of 12 units or >1 kg of field sample, e.g., potato mbers, cabbages, etc. Samples selected should not be damaged or suffer from severe defects, disease symptoms, or other abnormalities. 

Modifiers can be used very effectively in on-line SFE-GC to determine the concentration levels of the respective analytes. This presents an advantage in terms of the use of modifiers in SFE, since they appear as solvent peaks in GC separations and do not interfere with the target analyte determination. Although online SFE-GC is a simple technique, its applicability to real-life samples is limited compared to off-line SFE-GC. As a result, on-line SFE-GC requires suitable sample selection and appropriate setting of extraction conditions. If the goal is to determine the profile or matrix composition of a sample, it is required to use the fluid at the maximum solubility. For trace analysis it is best to choose a condition that separates the analytes from the matrix without interference. However, present SFE-GC techniques are not useful for samples... 

The determination of DDT residue on apples grown in the Hood River fruit district and on pears at Medford is carried on in branch laboratories established in those areas. The majority of samples selected for analyses are suspected of carrying higher amounts of residue than the average because of the spray program used or because the last application of insecticide was made within a few weeks of harvest. As indicated by Table I, about 80% of all the samples analyzed carried 3.0 p.p.m. or less of DDT during the past harvest season. Only about 20% of the samples showed residues above 3.0 p.p.m. six samples showed residue deposits slightly above 7.0 p.p.m. 

The substrate was Valencia orange leaves, with 2500 leaves per sample selected in a carefully prescribed manner (4). The trees involved were field sprayed in a conventional manner with 4 pounds of a 25% wettable powder of parathion, then sampled after 7 days and again after 11 days. Each sample was mixed thoroughly and subsampled into 125-leaf units in 2-quart Mason jars. To all units were added 250 ml. of benzene each, and they were sealed, stripped for various lengths of time, then restripped with fresh benzene, again for various lengths of time. The strip solutions were analyzed in the usual manner. 

The observations represent a sample selected randomly from a population which has been specified. 

This section is not intended to include sample selection, which will probably feature in separate sample plan documentation. Include sufficient detail to describe how the test portion is obtained from the sample received by the laboratory. Include storage, conditioning and disposal details. 

Indicator and sample selection are not the only choices a researcher has to make when using MAXCOV. A decision also has to be made about interval size, that is, how finely the input variable will be cut. Sometimes it is possible to use raw scores as intervals that is, each interval corresponds to one unit of raw score (e.g., the first interval includes cases that score one on anhedonia, the second interval includes cases that score two). This is what we used in the depression example. This approach usually works when indicators are fairly short and the sample size is very large, since it would allow for a sufficient number of cases with each raw score. In our opinion, this is the most defensible method of interval selection and should be used whenever possible. However, research data usually do not fit the requirements of this approach (e.g., the sample size is too small). Instead, the investigator can standardize indicators and make cuts at a fixed distance from each other (e.g.,. 25 SD), thereby producing intervals that encompass a few raw scores. 

The Waller et al. taxonic results do not appear to reflect a sample selection factor. If the findings resulted from combining predefined groups (individuals with multiple personality disorder and nonclinical controls), base rate estimates should have been about. 50. The. 37 taxon base rate suggests... 

It is important to note that the base rate estimates were very close to. 50, suggesting that the results could have simply reflected sample selection. One way to rule out this type of possibility is to compute Bayesian probabilities and show that not all taxon members come from the clinical group. Bayesian probabilities cannot be computed with SSMAXCOV, but MAXCOV can do it. The empirical indicators were probably too short for MAXCOV, but it may have worked with the theoretical indicators, as they are much longer. Unfortunately this analysis was not performed. 

A second variant detection strategy involves directed resequencing of selected genomic regions from different individuals, usually between 50 and 100 unrelated individuals. Many of these efforts have screened a subset of the DNA Polymorphism Discovery Resource collection, a set of 450 samples selected by the NIH to be a sample of the ethnic diversity of the U.S. popu-... 

In practice, from the analysis of the MALDI spectrum, it can be concluded that the sample selected is a mixture of linear and cyclic chains and that the linear chains are terminated in three different ways. 

Speed Useful for a variety of samples Selective and efficient columns High flow rate Fast data output Variety of detectors and stationary phases Low dead-volume fittings High-pressure pumps Fast-response recorders and automatic data handling... 

Fig. 1.9 [ 5 N. HJ-HSQC spectrum of 15N-uniformly labeled SH3 domain (A) and of a sample selectively...

Interestingly, in HPLC the stationary phase and the mobile-phase is able to interact with the sample selectively. Besides, such interactions as hydrogen bonding or complexation which are absolutely not possible in the GC-mobile phase may be accomplished with much ease in the HPLC-mobile phase. Furthermore, the spectrum of these selective interactions may also be enhanced by an appropriate chemical modification of the silica surface the stationary phase. Therefore, HPLC is regarded as a more versatile technique than GC and capable of achieving more difficult separations. 

Whereas precision (Section 6.5) measures the reproducibility of data from replicate analyses, the accuracy (Section 6.4) of a test estimates how accurate the data are, that is, how close the data would represent probable true values or how accurate the analytical procedure is to giving results that may be close to true values. Precision and accuracy are both measured on one or more samples selected at random for analysis from a given batch of samples. The precision of analysis is usually determined by running duplicate or replicate tests on one of the samples in a given batch of samples. It is expressed statistically as standard deviation, relative standard deviation (RSD), coefficient of variance (CV), standard error of the mean (M), and relative percent difference (RPD). 

---