K. Jagiełło et al.
16
visu alisation of multivariate data set; searching the relationships between sets of
data; recognition of internal data structure and making the classification [2]. Each
of them needs adequate type of data sets and usage the appropriate chemometric
approaches.
the next step is to design carefully the experiment and then perform measurements. the choice of the experimental design depends on the type of information
required to verify the research hypothesis, which in fact is the solution of the
defined problem. It is indicated to perform as little trials as possible, but selected
samples should be representative [1, 2]. In such way, we do not only save time
and cost of the whole project but also let make to increase significantly the quality
of the results.
Collected data should undergo multi-aspect control procedures to: (i) identify
the outliers; (ii) determine of the variable distribution that in some cases indicate
the necessity to perform data transformation; and, (iii) determine of interdependences between variables. the main goal of this procedure is to eliminate errors and
shorten the time necessary to get reliable results [2].
only data prepared according to described above procedures can be analysed
by means of chemometric techniques. the main approaches of chemometrics data
analysis are presented in Fig. 2.1.
Summing up, chemometrics is the field of science dealing with extracting
valuable information from multidimensional data sets by means of mathematematical and statistical methods. however, appropriate usage of chemometrics requires thinking about its application in each step of experiment: from the
problem defining to the analysis of obtained results.
2.2 Chemometrics and Numerical Approach
to a Set of Spectra
one of a typical problem in absorption spectroscopy is to determine an influence of
experimental conditions, such as temperature, ionic strength of the medium or ph,
on the electronic spectrum of tested compounds. the experimental condition may
affect intensity as well as shape of the spectrum. to detect such changes a set of
spectra registered at different experimental condition should be analysed.
From chemometrics point of view each spectrum could be treated as a onedimensional matrix. Absorptions at different wavelengths are elements of this
matrix. A set of spectra registered at different condition forms two-dimensional
matrix X [3, 4].
2.2.1 Preprocessing of Spectra
Registered sets of spectra, transformed to molar extinction scale according to the
Lambert-Beer Eq. (2.1) have to be preprocessed.
16
visu alisation of multivariate data set; searching the relationships between sets of
data; recognition of internal data structure and making the classification [2]. Each
of them needs adequate type of data sets and usage the appropriate chemometric
approaches.
the next step is to design carefully the experiment and then perform measurements. the choice of the experimental design depends on the type of information
required to verify the research hypothesis, which in fact is the solution of the
defined problem. It is indicated to perform as little trials as possible, but selected
samples should be representative [1, 2]. In such way, we do not only save time
and cost of the whole project but also let make to increase significantly the quality
of the results.
Collected data should undergo multi-aspect control procedures to: (i) identify
the outliers; (ii) determine of the variable distribution that in some cases indicate
the necessity to perform data transformation; and, (iii) determine of interdependences between variables. the main goal of this procedure is to eliminate errors and
shorten the time necessary to get reliable results [2].
only data prepared according to described above procedures can be analysed
by means of chemometric techniques. the main approaches of chemometrics data
analysis are presented in Fig. 2.1.
Summing up, chemometrics is the field of science dealing with extracting
valuable information from multidimensional data sets by means of mathematematical and statistical methods. however, appropriate usage of chemometrics requires thinking about its application in each step of experiment: from the
problem defining to the analysis of obtained results.
2.2 Chemometrics and Numerical Approach
to a Set of Spectra
one of a typical problem in absorption spectroscopy is to determine an influence of
experimental conditions, such as temperature, ionic strength of the medium or ph,
on the electronic spectrum of tested compounds. the experimental condition may
affect intensity as well as shape of the spectrum. to detect such changes a set of
spectra registered at different experimental condition should be analysed.
From chemometrics point of view each spectrum could be treated as a onedimensional matrix. Absorptions at different wavelengths are elements of this
matrix. A set of spectra registered at different condition forms two-dimensional
matrix X [3, 4].
2.2.1 Preprocessing of Spectra
Registered sets of spectra, transformed to molar extinction scale according to the
Lambert-Beer Eq. (2.1) have to be preprocessed.