Data pretreatment procedures

Moreda-Pineiro, A. Marcos, A. Fisher, A. Hill, S.J. Evaluation of the Effect of Data Pretreatment Procedures on Classical Pattern Recognition and Principal Components Analysis A Case Study for the Geographical Classification of Tea, J. Environ. Monit. 3(4), 352-360 (2001). 

We have found that rigorous sorbent pretreatment procedures (e.g., Soxhlet extraction and thermal desorption) in concert with a well-established quality control program will successfully control potential contamination effects arising from the sample collection media. Furthermore, a well-executed quality control program will permit identification of spurious data points attributable to media contamination when and if they do occur. 

The modelling of TPR patterns has not received much attention. Nevertheless, the results depend on the experimental conditions and modelling will be necessary when the results of different authors have to be compared. Moreover, the kinetic data obtained are useful in optimising pretreatment procedures. We present a convenient method of evaluating kinetic data from TPR patterns. 

It is necessary to acknowledge that some existing experimental data indicates that oxygen can be removed from Li/MgO directly during the reduction in hydrogen at temperatures as low as 873 K (see, for instance, Leveies, 2002). Such discrepancies with the data described above might be due to some difference in catalyst preparation/pretreatment procedures, which leads to the formation of active sites with somewhat different thermochemical characteristics. What is important is to attribute the evaluated kinetic parameters to the catalysts of particular thermochemistry. 

This application includes the task of identification and quantification of the adulterant. To demonstrate the different options in multivariate data analysis and achieve the aim of the study, we investigated the adulteration of olive oil by corn oil, soya oil, sunflower oil, walnut oil, and hazelnut oil. The NIR profiles of the samples of olive oil and the adulterated mixtures were used as the analytical data in this application. The measured infrared spectra were subjected to several different spectral pretreatment procedures, including double-derivative techniques, to identify the best... 

The Unscrambler (version 7.08, CAMO AS, Trondheim, Norway) software program was used in the PLS calibration of the NIR profiles with the adulterant percentage as the dependent variable in limited spectral regions and different data pretreatment methods. All of these different procedures were evaluated to find the spectral region and the data pretreatment method that give the best prediction and classification. The following procedure was found to serve the purpose of quantification and classification of an adulterant in olive oil in an acceptable manner. 

In particular in the industrial environment, where most in-poIymer analyses are being carried out, analyst time needs to be minimised. To this extent, autosamplers, robots, fast analysis techniques (e.g. ASE , fast GC), hyphenation and standardised data output formats for further manipulation and transmission are wanted. Automation is advantageous (Table 8.4). Ideally, the whole process may be automated analysis, data reduction and output. Unfortunately, standardisation of data handling procedures is still far off. This determines continuous, multiple efforts for training of analysts. Analytical methods should also be easy to maintain caUbra-tion should be required at minimal levels. Sample preparation should minimise time, effort, materials and volume of sample consumed. Sample pretreatment is ideally superfluous. There should also be little inherent doubt on the representativity of the analysis (of special concern for those techniques employing minimal sample amounts 0.1-1 fj,g). The method should be able to qualitatively identify the specific analyte(s) of interest, on the basis of expected behaviour (e.g. retention time, colour... 

Alternatively, data points can be collected in the decreasing pressure mode . This procedure is usually applied for the quantification of activated adsorption processes (Reuel and Bartholomew, 1984), such as the adsorption of H2. After the pretreatment of the sample (usually after reduction or reaction, and evacuation for a certain period to remove all the adsorbed surface species) the temperature is lowered to the temperature of measurement. First, a known amount of adsorbate gas is added to the reactor. Subsequently, the pressure in the catalyst compartment is lowered stepwise by expansion of the gas into the repeatedly evacuated reference volume. The adsorbed amount of gas can be calculated for each step. From this procedure, the monolayer capacity of the catalyst can be evaluated. 

In the following sections, common features of gas chromatographic procedures applied to proteinaceous material identification in paint are discussed such as sample pretreatments and data analysis. Finally, a section is devoted to the recognition of the amino acid racemisation in ancient proteins encountered mostly in archaeological contexts. 

Automatic analysis consists essentially of the same steps as the corresponding manual method (p. 4). In some cases this may be simple, the requirements amounting to a mechanical device for presenting the sample to the detector, a timer to control the time of measurement and a data recorder. However, if sample pretreatment and separations are necessary a variety of wet chemical stages needs to be automated. Such automated steps may be included in what remains essentially as an operator procedure. For... 

Analytical procedure is a systems problem and the samphng, pretreatment, measurement, data collection and reduction, and final reporting all have to be considered in a fiilly automatic approach. Computerization is often considered to he synonymous with automation but, although microprocessor technology is certainly changing the face of automatic instrumentation and influences both the control aspects and the data reduction, computerization is only a part of automation. Computers should simply be considered as tools of the trade within the area of automation. 

---