ThreeMode Abstracts, Part G
With one can go to the index of
this part of the Abstracts, with
one can go to other
parts (letters) of the Abstracts.
INDEX
Ga  Gb 
Gc  Gd 
Ge  Gf 
Gg  Gh 
Gi  Gj 
Gk  Gl 
Gm  Gn 
Go  Gp 
Gq  Gr 
Gs  Gt 
Gu  Gv 
Gw  Gx 
Gy  Gz 


Gabrielsson, A. (1979).
Dimension analyses of perceived sound quality of
soundreproducing systems. Scandinavian Journal of
Psychology, 20, 159169.
Perceived sound quality of loudspeakers, headphones, and hearing aids
was investigated by
different multivariate techniques to find out dimensions in perceived
sound quality and
explore their relations to physical characteristics of the systems. In
connection with the
description of two of these experiments various methodological questions
concerning judgment
methods, data treatment, interpretation of results, and implications for
continued research
are discussed. The combined results of eight experiments suggest the
following dimensions:
"Clearness/Distinctness", "Sharpness/Hardness  Softness", "Brightness 
Darkness",
"FullnessThinness", "Feeling of space", "Nearness", "Disturbing
sounds", and "Loudness".

Gabrielsson, A. & Sjögren, H. (1974, 1975).
Adjective ratings and dimension analyses of perceived sound
quality of hearing aids. I & II.
(Report TA No. 75 & 77). Stockholm: Karolinska Institutet, Technical
Audiology.
In the first report 6 program sections reproduced via 5 sound
reproducing systems (hearing aids) were scored on 62 tenpoint
adjective scales with respect to their sound quality (second
report: 6x8x40). In both studies T3 was applied with various
scalings and rotations, and the resulting factor spaces were
in many respects very similar to those in other factor
analyses, but not always so. The interpretation of the core
matrices was found to be too difficult. T3 was abandoned in
favour of INDSCAL.

Gabrielsson, A., & Sjögren, H. (1979a).
Perceived sound quality of hearing aids. Scandinavian
Audiology, 8, 159169.
Four experiments dealing with perceived sound quality of hearing aids are
described. The
purposes were (a) to find a limited number of perceptual dimensions
suitable for describing
variations in perceived sound quality, (b) to find the positions of the
selected hearing aids
in these dimensions, (c) to explore the relations between the perceptual
dimensions and the
physical characteristics of the hearing aids. Using various multivariate
techniques for
dimension analysis, the resulting dimensions were interpreted as
'Sharpness/Hardness 
Softness', 'Clearness/Distinctness', 'Brightness  Darkness', 'Fullness',
'Feeling of space',
'Nearness', 'Loudness', and 'Disturbing sounds'. Their relations to
characteristics of the
heaing aids were explored as well as their relations to overall
evaluations. The implications
for continued research and audiological applications are
discussed.

Gabrielsson, A., & Sjögren, H. (1979b).
Perceived sound quality of soundreproducing systems. Journal
of the Acoustical Society of America, 65, 10191033.
Perceived sound quality of loudspeakers, headphones, and hearing aids was
invesitgated by multivariate techniques from experimental psychology with the purpose (a)
to find out and interpret the meaning of relevant dimensions in perceived sound quality,
(b) find out the positions of the investigated systems in these dimension, (c) explore the
relations between the perceptual dimensions and the physical characteristics of the
systems, and (d) explore the relations between the perceptual dimensions and overall evaluations
of the systems. The resulting dimensions were interpreted as "clearness/distinctness,"
"sharpness/hardnesssoftness," "brightnessdarkness," "fullnessthinness," "feeling of
space," "nearness," "disturbing sounds," and "loudness." Their relations to physical
variables were explored by studying the positions of the investigated systems in the
respective dimensions. Their relations to overall evaluations were studied, and the
implications of the investigations for continued research are discussed.

Gailhard, I., Gros, P., Durbec, J. P., Beliaeff, B., Belin, C., Nezan, E., & Lassus, P. (2002).
Variability patterns of microphytoplankton communities along the French coasts.
Marine EcologyProgress Series, 242, 3950.
Microalgal populations along French coasts (English Channel, Bay of Biscay and
Mediterranean Sea) have been sampled twice a month since 1987 within the context of the French
Phytoplankton and Phycotoxin Monitoring Network (REPHY). This study used these data to
characterize the largescale geographical structures of microphytoplankton communities and to
determine whether 'homogeneous' geographical areas exist in which microalgal populations
display similar temporal variability schemes. Once the temporal variability component shared
by all sampled coastal sites was identified, the 'residual' sitespecific component was
analyzed. Multivariate ordination methods were used to determine seasonal and interannual
variability. The expectedtemporal pattern common to all sites was identified and the seasonal
cycle of the most frequently observed phytoplankton communities along French coasts was
described. The betweensite analysis, using multitable comparison methods (RVcoefficient and
multidimensional scaling), allowed the identification of 3 large areas (western English Channel,
Bay of Biscay and Mediterranean Sea) according to the temporal variability patterns of
microphytoplankton populations. The results, despite the coastal locations of REPHY sampling
sites, indicate that the hydrodynamic characteristics of the different areas play a major role
in the geographical structure of microalgal populations in French coastal waters.

Galera, M. M., Vidal, J. L. M., Frenich, A. G., & Gracia, M. D. G. (1997).
Evaluation of multiwavelength chromatograms for the quantification of
mixtures of pesticides by highperformance liquid chromatography diode
array detection with multivariate calibration.
Journal of Chromatography, 778, 139149.
Three multivariate calibration methods, partial least squares (PLS1 and PLS2) and principal
component regression, were applied to the simultaneous determination of the five pesticides
iprodione, procymidone, chlorothalonil, folpet and triazophos by highperformance liquid
chromatography with diode array detection. Such detection gives multiwavelength chromatograms from
a single analysis of one sample. In this paper,calibration models at two different
wavelengths were developed to resolve mixtures of five pesticides with overlapping chromatographic
peaks. The first model, carried out at 220 nm as detector compromise wavelength, yielded
satisfactory sensitivity for accurate estimation of the concentration of iprodione, procymidone,
chlorothalonil acid folpet and the second model, at 200 nm, was used for accurate estimation of
triazophos. Both calibration models were evaluated using the chromatograms and firstderivative (D1)
chromatograms by predicting the concentrations of independent test set samples. Finally, the
proposed D1 calibration models were successfully applied to the determination of these pesticides
in groundwater and soil samples. In all cases, the PLS1 calibration method showed superior
quantitative prediction ability than the PLS2 or principal component regression methods.

Gallagher, N. B., Shaver, J. M., Martin, E. B., Morris, J., Wise, B. M., & Windig, W. (2004).
Curve resolution for multivariate images with applications to TOFSIMS and Raman
Chemometrics and Intelligent Laboratory Systems, 73, 105117.
Multivariate curve resolution (MCR) is a powerful technique for extracting
chemical information from multivariate images (MI). Two problems with MI are (1)
initializing the MCR decomposition and (2) lack of selectivity in the image. Methods
derived for initializing MCR with evolving data that are naturally ordered in time are
not generally applicable for MI. Puritybased methods show promise and a simple, robust
puritybased algorithm has been developed to initialize the MCR decomposition. This
method used distance measures to find samples (or variables) on the exterior of a data set.
Lack of selectivity, common in MI, generally results in a rotational ambiguity in factors
extracted with MCR. Functional constraints were tested as a means to reduce this ambiguity,
and the method tested showed that functional constraints could be used to account for
offsets and backgrounds in Raman images. Robust initialization and introduction of
functional constraints tested here are necessary steps towards the final objective of
providing a simple methodology for constraining factors in a general fashion so that
knowledge of the physics and chemistry can be easily incorporated in to any MCR
decomposition. Additionally, the use of a sequential decomposition method (sequential MCR)
is employed to help reduce mixing of recovered components in rotationally ambiguous
systems.

Gandour, J.T. (1978).
Perceived dimensions of 13 tones: A multidimensional scaling
investigation. Phonetica, 35, 169179.
24 native speakers of American English made direct ratings of
dissimilarity
between 13 pitch patterns superimposed on a synthetic speechlike
syllable. A
multidimensional scaling analysis of the data revealed four perceptual
dimensions,
which weer interpreted as average pitch, endpoint, extreme endpoint and
length.
The relative importance of these dimensions varied across individual
subjects.

Gandour, J.T., & Harshman, R.A. (1978).
Crosslanguage differences in tone perception: A multidimensional scaling investigation.
Language and Speech, 21, 133.
Using an individual differences multidimensional scaling model of perception, this
crosslanguage investigation seeks to determine what dimensions underlie the perception
of linguistic tone, and to what extent an individual's language background (Thai,
Yoruba or American English) influences his perception. Dissimilarities data were obtained
from subjects' pairedcomparison judgments of 13 different pitch patterns superimposed
on a synthetic speechlike syllable. A multidimensional scaling analysis of the data
for the total group revealed that five dimensions  interpretively labeled, AVERAGE PITCH,
DIRECTION, LENGTH, EXTREME ENDPOINT and SLOPE  best summarize the perceptual structure
underlying the dissimilatities data. Language subgroup variation in relative importance
of these dimensions appears to be primarily related to subgroup differences in the
way pitch is used to convey linguistic information. Discriminant analysis showed that
most individual speakers of a tone language (Thai or Yoruba) can be easily distinguished
from speakers of a nontone language (English) on the basis of their distinctive patterns
of perceptual saliency for these five dimensions. Regression analysis indicated that
the DIRECTION and SLOPE dimensions closely correspond to certain earlier proposed binary
distinctive features of tone.

Gao, L., & Rhen, S. (1999).
Simultaneous spectrophotometric determination of four metals by the kernel partial
least squares method.
Chemometrics and Intelligent Laboratory Systems, 45, 8793.
Simultaneous determination of Mn, Zn, Co and Cd was studied by two methods, classical partial
least squares (PLS) and kernel partial least squares (KPLS), with 2(5bromo2pyridylazo)5diethylaminephenol
(5BrPADAP) and cetyl pyridinium bromide (CPB). Two programs, SPGRPLS and SPGRKPLS, were designed to perform
the calculations. Eight error functions were calculated for deducing the number of factors. Data reductions
were performed using principal component analysis. The KPLS method was applied for the rapid determination
from a data matrix with many wavelengths and fewer samples. Experimental results showed both methods to be
successful even where there was severe overlap of spectra.

García, I., Sarabia, L., Ortiz, M. C., & Aldama, L. M. (2004a).
Building robust calibration models for the analysis of estrogens by gas chromatography
with mass spectrometry detection.
Analytica Chimica Acta, 526, 139146.
Five hormonal growth promotants (diethylstilbestrol, hexestrol,
dienestrol, 17betaestradiol and 17alphaethynylestradiol) have been analysed by
gas chromatography with mass spectrometry detection (GC/MS, SIM mode) for four
nonconsecutive days. The aim is to build models with stable predictions. The strategies
applied are internal standardization and global models carried out by gathering signals
recorded on several days. Two models were examined: univariate models (with standardized
peak area) and PARAFAC2 (the analyte scores were standardized by the scores of the
internal standard). Internal standardization has been proved to be efficient for both
models of dienestrol and ethynylestradiol. The mean relative error in absolute value
when samples recorded on a different day to the calibration set are quantified by
PARAFAC2 is 7.00% and 7.11% for dienestrol and ethynylestradiol, respectively. For
diethylstilbestrol and estradiol, internal standardization was combined with global
calibration models built with signals recorded under several sources of variability
(different days). Thus predictions become steadier over time and in the estradiol
example, errors decrease from 33.10% to 9.76%. The mean relative error in absolute
value with PARAFAC2 updated models oscillates between 6.34% for ethynylestradiol and
10.74% for diethylstilbestrol. For univariate updated models errors range from 6.42%
to 14.19% for ethynylestradiol and estradiol respectively. The combination of both
strategies has been proved to be efficient independently of the analyte and of the
signal order.

García, I., Sarabia, L. A., & Ortiz, M. C.(2004b).
Detection capability of tetracyclines analysed by a fluorescence technique: comparison
between bilinear and trilinear partial least squares models.
Analytica Chimica Acta, 501, 193203.
According to the committee decision of 12 August 2002 (2002/657/EC) the capability
of detection, CCbeta, must be set in all analytical methods not only at concentration
levels close to zero but also at the maximum permitted limit (PL). In this work we
describe a methodology which evaluates the capability of detection of a fluorescence
technique with soft calibration models (bilinear and trilinear PLS) to determine
tetracyclines (group B1 substances from annex 1 of Directive 96/23/EC). Its estimation
is based on the generalisation of the procedure described in International Union of
Pure and Applied Chemistry and in the ISO standard 11843 for univariate signals
which evaluates the probabilities of false positive (alpha) and false negative (beta).
The capability of detection, CCbeta, estimated from the secondorder signal and the
trilinear PLS model is 9.93 mug l(1) of tetracycline, 17.75 mug l(1) of oxytetracycline
and 26.31 mug l(1) of chlortetracycline, setting alpha and beta at 0.05. The capability
of detection, CCbeta, determined around the PL (100 mug kg(1) in milk and muscle) with
the secondorder signal is 109.4 mug l(1) of tetracycline, 117.0 mug l(1) of
oxytetracycline and 124.9 mug l(1) of chlortetracycline, setting alpha and beta at 0.05.
The results were compared with those obtained with zero and firstorder signals. The
effect of the interferences on the capability of detection was also analysed as well
as the number of standards used to build the models and their calibration range. When
a tetracycline is quantified in presence of uncalibrated ones by means of the trilinear
PLS model the errors oscillate between 14.70% for TC and 9.57% for OTC.

García, I., Sarabia, L., Ortiz, M. C., & Aldama, L. M. (2004c).
Threeway models and detection capability of a gas chromatographymass spectrometry method
for the determination of clenbuterol in several biological matrices:
the 2002/657/EC European Decision.
Analytica Chimica Acta, 515, 5563.
Clenbuterol has been extracted by mixed solidphase extraction from two
biological matrices (bovine hair and urine) and detected by GC/MS (selected ion monitoring
(SIM) and fullSCAN modes). The analytical signal has been modelled with univariate and
threeway models, namely DTLD, PARAFAC, PARAFAC2, Tucker3 and trilinear PLS. Since
clenbuterol is a banned substance a comparative study of the capability of detection
(CCbeta, X0 = 0) has been performed as a function of the sample (hair, 74 mug kg(1) and
urine, 0.36 mug l(1)), the mode in which the signals are monitored (SCAN, 283 mug kg(1)
and SIM, 74 mug kg(1)) and the statistical model (univariate, 283 mug kg(1) and trilinear
PLS, 20.91 mug kg(1)). The capability of detection has been calculated as stated in
ISO 11843 and Decision 2002/657/EC setting in all cases the probabilities of false positive
and of false negative at 0.05.
The identification of the mass spectra must be done to confirm the presence of clenbuterol
and has been carried out through PARAFAC. The correlation coefficient between the spectra
estimated by PARAFAC and the library spectra is 0.96 (hair, SCAN mode) and 1.00 (hair and
urine, SIM mode).
The Decision 2002/657/EC advocates the use of independent mass fragments to identify banned
compounds. These recommendations together with the effect of the number of ions registered
on the capability of detection have lead us to select five uncorrelated fragments
(86, 243, 262, 264 and 277) from the data set of 210 ions by hierarchical clustering of
variables.

García, J.M., Jimenez, A.I., Arias, J.J., Khalaf, K.D., Moralesrubio, A.,
& Delaguardia, M. (1995).
Application of the partial leastsquares calibration method to the simultaneous kinetic
determination of propoxur, carbaryl, ethiofencarb and formetanate. Analyst,
7, 7.
A method based on the application of partial leastsquares analysis to bilinear data
was used for the simultaneous determination of propoxur, carbaryl, ethiofencarb and
formetanate by a kineticspectrophotometric method. The procedure is based on the
different rate constants of the reactions between paminophenol, in the presence
of potassium metaperiodate, and the phenolic and naphtholic compounds obtained from
the alkaline hydrolysis of the pesticides using a stoppedflow injection procedure.
Mixtures containing 210 mu g ml(1) of propoxur, 28 mu g ml(1) of carbaryl,
210 mu g ml(1) of ethiofencarb and 210 mu g ml(1) of formetanate were successfully
resolved with errors of less than 5%.

García, J. A. J.., Plaza, J. G., & Pavon, J. M. C.(1994).
Resolution of overlapped peaks in liquidchromatography by use of derivative
spectrophotometry and multivariateanalysis and its application in the determination
of active components in insecticide formulations.
Fresenius Journal of Analytical Chemistry, 349, 542545.
Two procedures for the determination of active
components in insecticide formulations have been
devised. The first is based in the use of
derivative spectra of the components obtained by a
diodearray spectrophotomer around the maxima
signal of the chromatographic peak. In the second
method, mixtures are resolved by the partial
leastsquares (PLS) regression method from
standard spectra of the pure components; spectra
of the components were also registered around the
maxima signal of the peak. Both procedures have
been applied to the analysis of diverse mixtures
of active components (piperonyl butoxide,
neopynamine and fenitrothion) in insecticide
formulations with satisfactory results.

Gardner, W. P., Shaffer, R. E., Girard, J. E., & Callahan, J. H. (2001).
Application of quantitative chemometric analysis techniques to direct
sampling mass spectrometry.
Analytical Chemistry, 73, 596605.
This paper explores the use of direct sampling mass spectrometry
coupled with multivariate chemometric analysis techniques for the analysis of
sample mixtures containing analytes with similar mass spectra, Water samples
containing varying mixtures of toluene, ethyl benzene, and cumene were analyzed
by purgeandtrap/direct sampling mass spectrometry. Multivariate calibration
models were built using partial leastsquares regression (PLS), trilinear partial
leastsquares regression (triPLS), and parallel factor analysis (PARAFAC), with
the latter two methods taking advantage of the differences in the temporal profiles
of the analytes. The prediction errors for each model were compared to those
obtained with simple univariate regression. Multivariate quantitative methods
were found to be superior to univariate regression when a unique ion for
quantitation could not be found. For prediction samples that contained unmodeled,
interfering compounds, PARAFAC outperformed the other analysis methods. The
uniqueness of the PARAFAC model allows for estimation of the mass spectra of the
interfering compounds, which can be subsequently identified via visual inspection
or a library search.

Gargallo, R., Tauler, R., Sanchez, F., & Massart, D. L. (1996).
Validation of alternating leastsquares multivariate curve resolution for
chromatographic resolution and quantitation.
TracTrends in Analytical Chemistry, 15, 279286.
The application of multivariate curve resolution
using the alternating leastsquares (ALS) method
for the resolution and quantitation of mixtures of
two coeluted compounds is validated for
hydrocortisone (main compound) and
prednisone(impurity) in liquid chromatography with
diode array detection. The recovery of the spectra
of the coeluted compounds and their quantitation
are discussed in relation to the initial estimates
provided by evolving factor analysis and SIMPLISMA
methods. Quantitation errors below 1.5% have been
obtained for the main compound for resolution
values of 0.4 or above, and impurity
concentrations in the range 0.520%.

Geladi, P. (1989).
Analysis of multiway (multimode) data. Chemometrics and
Intelligent Laboratory Systems 7, 1130.
The concepts data matrix and multivariate data analysis are rapidly
becoming popular and wellknown words in chemistry. Many methods used in
the laboratory can produce data arrays of a greater complexity than the
data matrix. The broad picture easily gets lost here, not least because
of the need for systematization and generalization.

Geladi, P. (1992a).
Some special topics in multivariate imageanalysis.
Chemometrics and Intelligent Laboratory Systems, 14, 375390.
Methods for multivariate image analysis are based on the definition
of the multivariate image as a stack of congruent images collected for different
variables (wavelengths). The article describes theories and three different types
of examples for analysis when the images available are not congruent. This requires
transformation to a common base used to construct a multivariate image. After that,
a multivariate analysis can be carried out and the loading plots can be used for
exploratory analysis and classification. Two new strategies are introduced: one for
comparing within a set of noncongruent univariate images and one for comparing
within a set of multivariate images.

Geladi, P. (1995).
Sampling and Local Models for Multivriate Image Analysis. Mikrochimica
Acta120, 211230.
Local models are a very important concept for microscopic and macroscopic
imaging. Different methods of subsampling a multivariate image are described
both in general and for three examples. The need for subsampling and its
influence on multivariate image analysis and visualization are studied.
Examples from MRI (256×256), satellite imaging (7 × 512 ×
512) and biofuel studies (6 × 512 × 512)are used to illustrate
som of the principles involved.

Geladi, P., & Aberg, P. (2001).
Threeway modelling of a batch organic synthesis process monitored by near infrared spectroscopy.
Journal of Near Infrared Spectroscopy, 9, 19.
Multivariate monitoring by near infrared (NIR) spectroscopy of an
organic synthesis as a batch process gives threeway data of the type batch number
x NIR wavelength x time, The data can be analysed by threeway decomposition methods
such as parallel factor analysis (PARAFAC). The batch process is the synthesis of an
ester carried out by students in the laboratory. The synthesis is monitored by
fibreoptic probe NIR measurement every ten minutes. The NIR spectra obtained have no
distinct peaks and simply doing curve resolution on them is very difficult, but it is
shown that pretreatment of the NIR spectra can lead to a good PARAFAC model and a
chemical interpretation of the unique loadings obtained. The interpretation of the
spectral loadings is carried out by comparison with pure chemical spectra. The data
set is available from the authors.

Geladi, P., Bergner, H. & Ringqvist, L. (2000).
From experimental design to images to particle
size histograms to multiway analysis. An example
of peat dewatering.
Journal of Chemometrics, 14, 197211.
The efficiency of peat dewatering by filtering
slurries is dependent on the sizes of fine and
colloidal particles that clog the filter. A
designed experiment was carried out to check the
use of different treatments on particle
coagulation. The resulting particle sizes were
studied under the microscope by automated digital
image analysis, leading to area histograms for 21
size classes. Seven treatments on five peat types
give a twoway ANOVA in allqualitative
variables, but the 21 response variables are a
bit too much for an ANOVA or MANOVA analysis. The
data can also be arranged in a 5 (peat types) x 7
(treatments) x 21 (size classes) threeway array.
This array is analyzed by parafac and gives an
effective threeway rank of 4. The threeway data
have no obvious underlying trilinear structure,
and curve resolution results are not expected.
The threeway analysis gives a very parsimonious
model that is easily interpreted as a function of
the problem definition. The emphasis is on
visualization of the results.

Geladi, P., & Forsstrom, J. (2002).
Monitoring, of a batch organic synthesis by nearinfrared spectroscopy: modeling
and interpretation of threeway data
Journal of Chemometrics, 329338.
Threeway data of the type batch x time x NIR wavelength were obtained by NIR
spectroscopic multivariate monitoring of an organic synthesis as a batch process.
The model synthesis, an ester synthesis, was carried out as an experimental design.
Unexpected technical problems caused a blocking effect that forced a modification
of the design. After preprocessing of a reduced threeway array, the spectral data
in the threeway array were subjected to parallel factor analysis (PARAFAC). The
loadings from this analysis could be interpreted and explained as a function of
the synthesis studied. For the spectral interpretation, spectra of pure chemicals
were needed. The paper is an illustration of what can be done with threeway modeling
in order to increase the understanding of a reaction, and it attempts to show how
the results can be interpreted and presented. The data sets are available from the
authors.

Geladi, P., Bengtsson, E., Esbensen, K., & Grahn, H. (1992b).
Imageanalysis in chemistry. I. Properties of images, graylevel operations,
the multivariate image.
TracTrends in Analytical Chemistry, 11, 4153.
The use of imaging and images in chemical and technological
applications is treated in two parts. Part I focuses on the generation and
properties of images. The result of image registration is an analog or digital
image, a representation of intensities as a function of position coordinates x,
y (and sometimes also z) in a nonhomogeneous material. Certain univariate
operations for reducing noise and errors, improving contrast, measuring and
counting particles etc. can be carried out on digitized images. A simple example
from bioenergy research is given. In this first part a multivariate image is
defined and an explanation is given of how it can be obtained. Operations on
multivariate images and interpretations of multivariate results are presented in
part II.

Geladi, P., Grahn, H., Esbensen, K., & Bengtsson, E. (1992c).
Imageanalysis in chemistry. II. Multivariate imageanalysis.
TracTrends in Analytical Chemistry, 11, 121130.
Part I [Trends Anal. Chem., 11 (1992) 41] focused on the generation
of univariate digital images in the chemical laboratory and in industrial
situations and on the possible use of operations on univariate images. The concept
of a multivariate image was introduced and an example given. This part focuses on
the use of multivariate methods to extract problemdependent, useful information
from multivariate images. The example of the powder mixtures given in Part I is
further analyzed by principal component analysis. Concepts of exploratory analysis,
classification and regression are explained. The use of visual interpretation and
statistical diagnostics is emphasized. Ideas about future developments of
multivariate image analysis are introduced.

Geladi, P., Manley, M.& Lestander, T. (2003)
Scatter plotting in multivariate data analysis
Journal of Chemometrics, 17 503511.
In data analysis, many situations arise where plotting and visualization are
helpful or an absolute requirement for understanding. There are many techniques
of plotting data/parameters/residuals. These have to be understood and visualization
has to be made clearly and interpreted correctly. In this paper the classical
favourites in chemometrics, scatter plots, are looked into more deeply and some
criticism based on recent literature references is formulated for situations of
principal component analysis, PARAFAC threeway analysis and regression by partial
least squares. Biplots are also afforded some attention. Examples from nearinfrared
spectroscopy are given as illustrations. Copyright (C) 2003 John Wiley Sons, Ltd.

Geladi, P.,Swerts, J. & Lindgren, F. (1994).
Multiwavelength microscopic image analysis of a piece of painted chinaware:
classification and regression.
Chemometrics and Intelligent Laboratory Systems, 24, 145167.
A multivariate microscopic study of a piece of painted china poreclain is
undertaken by image analysis. The object has archeological and artistic value
and therefore a detailed study may be worthwhile. The use of multivariate image
analysis intends the exploration of the artefact (artistic object) in more
spectral and spatial detail than by just visual inspection. The study is used
as a means of introducing and further exploring the different aspects of
multivariate image analysis. The goal of the paper is twofold: (1) showing how
the multivariate image is constructed and analyzed and (2) using some of the
obtained results to introduce and further expand some of the techniques of
multivariate image analysis. The example image has a size of 6 x 512 x 512.
Classifications by feature space segmentation and by regression are shown to
be useful and objective methods of acquiring insight in measurement errors and
in the artistic detail of the painting. Some new concepts for multivariate
analysis are introduced. One of them is related to the comparison of regression
models.

Geladi, P., Sethson, B., Nystrom, J., Lillhonga, T., Lestander, T. & Burger, J. (2004).
Chemometrics in spectroscopy  Part 2: Examples.
Spectrochemica Acta Part B Atomic Spectroscopy, 59 13471357.
Some of the principles and main methods of chemometrics are illustrated
by examples. The examples ate from electrochemistry, process analytical chemistry and
multivariate imaging. Principal component analysis, partial least squares regression
and multivariate image analysis are used to illustrate the power of chemometrical
thinking. The emphasis is on plotting and visualization for showing the salient features
of a model or data set.

Geladi, P., Xie, Y. L., & Hopke, P. (1998).
Regression on parameters from threeway decomposition.
Journal of Chemometrics, 12 337354.
This paper presents work on the combination of (a) environmental chemistry, (b)
threeway analysis by Parafac constrained to nonnegative results and (c) multivariate calibration.
For two different environmental examples it is shown how the loading parameters from an independent
threeway analysis can be used in a regression against external data. This combination leads to an
easier interpretation of the results. The data are from Arctic aerosol studies. The subjectively
obtained Parafac loadings are regressed against temperature anomalies.

Gemperline, P.J., Rulifson, R.A., & Paramore, L. (2002).
Multiway analysis of trace elements in fish otoliths to track migratory patterns.
Chemometrics and Intelligent Laboratory Systems, 60, 135146.
Spawning striped bass in the Shubenacadie watershed of Nova Scotia, Canada
exhibit three dorsal coloration patterns: green, indicative of fish from
the ocean; black, indicative of fish that overwinter in a fresh headwater lake,
and mottled fish of unknown origin. Microchemical analysis of growth rings
in fish otoliths (calcareous particles found in the inner car of certain
lower vertebrates), measured by laser ablationinductively coupled plasma/mass
spectrometry (LAICP/MS), from fish captured during the 1999 Shubenacadie
spawning period were analyzed by Tucker3 multiway principal component models.
Using this technique, multidimensional patterns were discovered in the trace
element measurements indicating that migratory patterns of individual striped
bass can be tracked from the timedependent trace element record deposited
in the otoliths. Of the nine fish analyzed by LAICP/MS, trace element composition
at year 0 suggested that all nine fish originated from the same locale.
Differentiation in the trace element record was observed in subsequent years.
Clustering of the trace element data for six fish unambiguously coincided with
dorsal coloration. The three remaining fish exhibited trace element patterns
that suggested migration between freshwater and marine conditions at one or
more periods during life.

Geng, C., & Mooshammer, C. (2000).
Modeling the German stress distinction. In Proceedings of the Fifth Seminar
on Speech Production: Models and Data (pp. 161164). Kloster Seeon
Germany, 14 May.
Low dimensional and speakerindependent linear vocal tract parametrications can
be obtained using the 3mode PARAFAC factor analysis procedure first introduced
by Harshman et al. (1977). The following study used PARAFAC to investigate the
stress distinction in German vowel production. Tongue movements of six German
speakers were recorded by means of EMMA. The data was entered into the classical
PARAFAC1 model treating the stress distinction for each subject as two different
speakers. This gave a reasonable 2factor solution, but was not without
drawbacks. The model turned out to be capable of recovering gross anatomical
properties of our subjects, but failed to return intraindividual differences in
tongue shapes with respect to word stress. This indicated that the strict
linearity assumptions required in the classical PARAFAC model were too strong to
capture stress specific variation in full detail. We supposed that a model
closely related to PARAFAC, PARAFAC2, should allow to account for systematic
variation produced by word stress by imposing weaker structure on the data. As
will be shown, PARAFAC2 modeled the physical properties of the vocal tract shape
in a more realistic and plausible way.

Gensch, D. H., & Javalgi, R. G. (1988).
Issues and advances in product positioning models in marketingresearch.
Mathematical and Computer Modelling, 10, 929949.
Mathematical models used for product positioning are proliferating at a rapid
rate. To provide a structure to this research, we have categorized the models/algorithms into
three basic approaches: utility functions, maps and trees. In this article we identify and
discuss the recent advances related to these approaches, discuss data requirements and indicate
issues that influence that choice of the particular product positioning approach bestsuited
for various situations.

Gervini, D., & Rousson, V. (2004).
Criteria for evaluating dimensionreducing components for multivariate data.
American Statistician, 58, 7276.
Principal components are the benchmark for linear dimension reduction, but
they are not always easy to interpret. For this reason, some alternatives have been proposed
in recent years. These methods produce components that, unlike principal components, are
correlated and/or have nonorthogonal loadings. This article shows that the criteria commonly
used to evaluate principal components are not adequate for evaluating such alternatives, and
proposes two new criteria that are more suitable for this purpose.

Gimeno, R. A., Beltran, J. L., Marce, R. M., & Borrul, F. (2000).
Determination of naphthalenesulfonates in water by online ionpair solidphase
extraction and ionpair liquid chromatography with fastscanning fluorescence
detection.
Journal of Chromatography A, 890, 289294.
A fast analytical method for quantifying a mixture of 12
naphthalenesulfonates and naphthalenedisulfonates has been developed. This method
consists of online ionpair solidphase extraction with PLRPs sorbent and ionpair
liquidchromatography using fastscanning fluorescence spectrometer as a detection
system and multivariate calibration. As complete separation is unnecessary, the
compounds were analysed in isocratic conditions and the chromatographic analysis
took only 25 min. Threeway partial leastsquares (PLS) was used to carry out
multivariate calibration for spiked tap water. in these conditions, quantification
limits were between 0.01 and 3 mu g l(1). Repeatability was also evaluated and
relative standard deviations (n = 3) were between 0.5 and 4, depending on the
compound. Finally, spiked tap and Ebro river waters were analysed to evaluate
prediction capability of the method.

Giordani, P., & Kiers, H. A. L. (2004).
Threeway component analysis of intervalvalued data.
Journal of Chemometrics, 18, 253264.
Vertices Principal Component Analysis (VPCA) and Centers Principal Component Analysis
(CPCA) are variants of Principal Component Analysis (PCA) to deal with twoway intervalvalued
data. In this case the observation units are represented as hyperrectangles instead of points. Tucker3
and CANDECOMP/PARAFAC are component analysis techniques to analyze the underlying
structure of threeway data sets. In the present paper, after recalling the above mentioned methods,
we extend the CPCA and VPCA methods to deal with threeway intervalvalued data by means of
Tucker3 and CANDECOMP/PARAFAC and we describe how to represent the observation units in
the obtained lowdimensional space. Furthermore, an application of the extended methods—called
Threeway Vertices Principal Component Analysis (3VPCA) and Threeway Centers Principal
Component Analysis (3CPCA)—to threeway intervalvalued air pollution data is described.

Gitin, S.R. (1970).
A dimensional analysis of manual expression.
Journal of Personality and Social Psychology, 15, 271
277.
T3 was performed to investigate 78 subjects' ratings of 36
photographs of manual expressions on 40 semanticdifferential
type scales. The expression on scale configurations were
varimax rotated, but the core matrix (= 1 core plane) is given
for the unrotated components. It shows a nice 'simple'
(diagonal) structure.

Gilbert, D.A., Sutherland, M., & Kroonenberg, P.M. (2000).
Exploring subjectrelated interactions in repeated measures data using three
mode principal components analysis. Nursing Research, 49, 57
61.
In this paper a case is made for analyzing subjectrelated interactions
in repeated measures data in some detail rather than stopping at
significance tests. In particular, such interactions can be usefully
modeled with multiplicative models, and threeway interactions with
threemode principal component models. The proposal is illustrated with
data on the judgments of relational communication by nurses.

Glasbey, C. A., & Glidewell, S. M. (2001).
The application of multivariate statistical methods to NMR imaging.
Computers and Electronicx in Agriculture, 32, 85100.
Statistical methods are used to integrate the information content
in multivariate NMR images which had been obtained using different instrumental
settings. Empirical and mechanistic models of multivariate pixel values are
compared, making use of a novel reparametrisation of the mechanistic parameters to
improve numerical stability. A new approach is presented to integrating the
information using colour image displays. Projection pursuit is used to identify
optimal instrumental settings in synthetic images, Methods are illustrated using
seven crosssectional NMR microscopy images of a blackberry, which was chosen as
typical of Fruit specimens with hard tissue distributed in a soft matrix.

Goodchild, M. F., Klinkenberg, B., & Janelle D. G. (1993).
A factorial model of aggregate spetiotemporal behavior  application to the
diurnal cycle.
Geographical Analysis, 25, 277294.
The crosssectional nature of much social data, coupled with
the static view provided by maps and current spatial data
handling software, have produced a tradition of research on
urban spatial structure that is largely twodimensional and
derived from residential locations. The paper presents an
analysis of a spacetime diary data set collected in Halifax,
Nova Scotia. A series of transformations are used to convert
the individual diary records to a threemode matrix of
intensities, which is then analyzed using the PARAFAC three
mode factor model Home / work is found to be the strongest
organizing dimension of the urban spacetime, followed by
entertainment, shopping, and education / work. We show how
these dimensions appear to varying degrees in different
locations, time periods, and human activities. The paper argues
for a dynamic view of urban spatial structure in which only the
physical facilities remain static.

Gower, J.C. (1975).
Generalised Procrustes analysis. Psychometrika, 40,
3351.
Suppose P_{j}^{(i)} (i =
1, 2, ... , m, j = 1, 2, ... , n) give the
locations of mn points in pdimensional space.
Collectively these may be regarded as m configurations, or
scalings, each of n points in pdimensions. The problem
is investigated of translating, rotating, reflecting and scaling the
m configurations to minimize the goodnessoffit criterion
Sigma_{i=1m}
Sigma_{j=1n} Delta²
(P_{j}^{(i)}
G_{j}), where G_{j} is the centroid of
the m points
P_{j}^{(i)} (i = 1,
2, ..., m). The rotated positions of each configuration may be
regarded as individual analyses with the centroid configuration
representing a consensus, and this relationship with individual
scaling analysis is discussed. A computational technique is given, the
results of which can be summarized in analysis of variance form. The
special case m = 2 corresponds to Classical Procrustes analysis
but the choice of criterion that fits each configuration to the common
centroid configuration avoids difficulties that arise when one set is
fitted to the other, regarded as fixed.

Gower, J.C. (1977).
The analysis of threeway grids. In P. Slater (Ed.), Dimensions
of Interpersonal Space (pp. 163173). New York: Wiley.
To analyse threeway data a set of models are proposed within the analysis
of variance context. In particular, different ways are given to decompose
the residuals after the main effects have been removed. Introducing twoway
interactions leads to tractable solutions of uncertain convergence speed,
but the inclusion of three way interaction terms introduced difficulties
to which no solutions were found. The position of INDSCAL within this
framework is discussed.

Gower, J.C. (1984).
Multidimensional scaling displays. In H.G. Law, C. W. Snyder Jr, J. A. Hattie &
R. P. McDonald (Eds.), Research methods for multimode data analysis (pp.
592601). New York: Praeger.
Graphical displays for various forms of threemode multidimensional scaling may
appear quite varied, but analysis shows that they are made up from a few basic
geometrical ideas. This paper attempts to uncover these basic displays  each of
which is usually appropriate to twomode forms of analysis  and shows how they
may be combined to give threemode displays. With multidimensional scaling, like
other multivariate methods, the most natural graphical displays directly reflect
the properties of the algebraic form of the model fitted. Thus, when a distance
matrix is fitted to data by multidimensional scaling, its simplest
representation is by a set of points such that the distance between them is, or
is proportional to their similarity. With an innerproduct model, it is natural
to plot vectors and the angle between pairs of vectors together with their
lengths is the geometrical property used for interpretation. Both distances and
innerproducts generate symmetric matrices. These basic plotting devices are
discussed in more detail in the article. The plots associated with threemode
arrays usually combine elements of the basic twodimensional displays. From the
above remarks, it should be clear that the form of plotting used cannot be
entirely divorced from considerations of the model fitted.

Gower, J.C. & De Rooij, M. (2003).
A Comparison of th Multidimensional Scaling of Triadic and Dyadic Distances.
Journal of Classification, 20, 115136.
We examine the use of triadic distances as a basis for multidimensional scaling (MDS).
The MDS of triadic distances (MDS3) and a conventional MDS of dyadic distances (MDS2)
both give Euclidean representations. Our analysis suggests that MDS2and MDS3 can
be expected to give very similar results, and this is strongly supported by
numerical examples. We have concentrated on the perimeter and generalized Euclidean
models of triadic distances, both of which are linear transformations of dyadic
distances and so might be suspected of explaining our findings: however an MDS3
of the nonlinear variance definition of triadic distance also closely approximated
the MDS2 representation. An appendix gives some matrix results that we have found
useful and also gives matrix representations and alternative derivations of some
known properties of triadic distances.

Grahn, H. F., & Saaf, J. (1992).
Multivariate image regression and analysis  useful techniques for the evaluation
of clinical magneticresonance image.
Chemometrics and Intelligent Laboratory Systems, 14, 391396.
Multivariate image analysis (MIA) and multivariate image regression
(MIR) techniques are useful tools in the extraction of information from magnetic
resonance images. They aid the characterization of different tissues and can be
used to describe their size and distribution. The obtained new information can be
used to monitor growth, progression and effects of a treatment. The methodology is
illustrated by a clinical example. The ongoing development of MIA and MIR,
combining parameters from different imaging modalities, is commented on.

Gräser, H. (1977).
Spontane Reversionsprozesse in der
Figuralwahrnehmung. Eine Untersuchung reversibler Figuren mit
der Dreimodalen Faktorenanalyse (doctoral thesis). Trier,
FRG: author.
The reversion rate for 46 reversible figure of 133 subjects was
determined at 15 time points. T3 as described by Bartussek (1973)
was used; all factor matrices were rotated by various methods.
Two different standardizations of the data were used;
the merits of both are discussed. Reification with external
criteria. Extended core matrices were derived by
premultiplying the core matrix with component matrices.
Interpretation of change factors in T3 is discussed. Includes
an Appendix with a program description for T3.

Gräser, H., Esser, H. & Saile, H. (1981).
Einschätzung von
Lebensereignissen und ihren Auswirkungen. In S.H. Filipp (Ed.)
Kritische Lebensereignisse und ihre
Bewältigung (pp. 104122). München, FRG: Urban &
Schwarzenberg.
In one of two reported studies on the perception
of life events 91 subjects scored 20 life events on 19 sentiments
and characteristics. The 18 scales were standardized over
subjects and events. Events and scales were varimaxed, and the
subjects were obliquely rotated. The core matrix, multiple
regression and discriminent analysis on external variables were
used to aid the interpretation of the subjects.

Gredelj, S., Gerson, A.R., Kumar, S., & Cavallaro, G.P. (2001).
Characterization of aluminium surfaces with and without plasma nitriding
by Xray photoelectron spectroscopy. Appliec Surface Science, 174, 240250.
Substrates of aluminium alloy 2011 have been plasma nitrided using an inductively
coupled plasma source and then characterised by Xray diffraction (XRD)
and Xray photoelectron spectroscopy (XPS). The XRD analysis confirmed the
presence of a crystalline aluminium nitride (AlN) layer on the substrates
surfaces. Both AlN and Al2O3 were observed by XPS on the surface of the nitrided
Al substrates. The binding energy (BE) of the C 1s photoelectron of
adventitious hydrocarbon was found to be dependent on the thickness of the
insulating AlN/Al2O3 surface layer. None of the Al 2p, N 1s or O 1s BEs showed
any shifts due to surface charge build up. The BE values of the Al 2p photoelectron
for AlN, Al2O3 and metallic Al were determined to be 74.5, 75.5 and 72.7 eV, respectively.
The charge shifting of the BEs of all the photoelectrons occurs simultaneously
only for oxidised Al substrates (not nitrided) with a sufficiently thick oxidised
surface layer. For a thinner oxidised surface layer the adventitious hydrocarbon
C 1s photoelectron BE still shows charge shifting but this is not reflected in
the BE of the Al 2p and O 1s XPS components.

Green, P.E., Carmone, F.J., & Wachspress, D.P. (1976).
Consumer segmentation via latent class analysis.
Journal of Consumer Research, 3, 170174.
A nontechnical description of latent class analysis and a discussion of
the CANDECOMP procedure and its application to consumer adoption of a new
telecommunications service is presented with some suggestions for other areas in
consumer and market research wheer latent class analysis might be of use.

Green, P. E., & Devita, M. T. (1974).
A complementarity model of consumer utility for item collections.
Journal of Consumer Research, 1, 5667.
A model is developed that portrays certain types of interactions as well
as main effects in consumer evaluation tasks. The model is illustrated with menu preference
data and its potential use in other types of multiattribhute choice situations is described.

Groenen, P. J. F., Van Os, B. J., & Meulman, J. J. (2000).
Optimal scaling by alternating lengthconstrained nonnegative least squares,
with application to distancebased analysis.
Psychometrika, 65, 511524.
An important feature of distancebased principal components
analysis, is that the variables can be optimally transformed. For monotone
spline transformation, a nonnegative leastsquares problem with a length
constraint has to be solved in each iteration. As an alternative algorithm to
Lawson and Hanson (1974), we propose the Alternating LengthConstrained
NonNegative LeastSquares (ALCNNLS) algorithm, which minimizes the nonnegative
leastsquares loss function over the parameters under a length constraint, by
alternatingly minimizing over one parameter while keeping the others fixed.
Several properties of the new algorithm are discussed. A Monte Carlo study is
presented which shows that for most cases in distancebased principal components
analysis, ALCNNLS performs as good as the method of Lawson and Hanson or
sometimes even better in terms of the quality of the solution.

Grotti, M. (2004).
Improving the analytical performances of inductively coupled plasma optical emission
spectrometry by multivariate analysis techniques.
Annali Di Chimica, 94, 115.
The various multivariate analysis techniques which have been successfully applied to
maximize the analytical performances of ICPOES are reviewed. These include optimization
procedures, spectral data processing and calibration methods as well as classification and
pattern recognition techniques.

Grotti, M., Rivaro, P., Leardi, R., & Magi, E. (1999).
Threeway principal component analysis as a powerful tool to process marine
environmental data sets.
Annali Di Chimica, 89, 591600.
A study of the marine ecosystem usually requires management of large and
complex data sets. The multivariate technique "Threeway principal component analysis"
has been applied to a relatively small set of oceanographic data. Sea water samples were
collected during the Italian Antarctic Expedition 199495 and the following parameters
were considered: sea water temperature and salinity, dissolved oxygen, nutrients
(silicate, phosphate, nitrate plus nitrite) and phytoplanktonic pigments (total
chlorophyll, chlorophyll a, phaeopigments, carotenoids). The resulting plots confirmed
in a selfexplanatory way some wellknown assumptions, thus showing that
"Threeway Principal Component Analysis" is a powerful instrument for the
interpretation of marine environmental data sets.

Groves, C.L. (1978).
Individual difference modelling of simple
functional relations: Examples using threemode factor
analysis. Dissertation Abstracts International,
39 (5B), 24752476.
Simulated fallible data were used to examine the influence of
several types of error on the ability of threemode factor
analysis to recover both the stimulus differences in simple
functional relations data and the individual differences.
Experimental data pertaining to the sizeweight illusion were
examined.

Grung, B., & Kvalheim, O.M. (1995a).
Detection and quantification of embedded minor analytes in
threeway multicomponent profiles by evolving projections and
internal rank annihilation. Chemometrics and Intelligent
Laboratory Systems, 29, 213221.
Modern analytical instrumentation often leads to data arrays with more than
two dimensions or directions. Such Nway data (N > 2)
needs special resolution methods for optimising the amount of analytical
information. In this work, a new interactive method designed to work with
threeway data is presented. The method, evolving projections by optimised
search (EPOS), presents a combined graphical and numerical way of resolving
a threeway data array into the analytical profiles of the pure analytes.
The method involves an internal rank annihilation step which can be
performed in several ways. The graphic interactive procedure used in this
work compares favourably with Lorber's noniterative rank annihilation method.
Thus, our method is significantly better for resolution of analytes with
low relative concentration, especially in the presence of heteroscedastic
noise. The EPOS method is tested on several simulated data sets to assess
its performance. A peak purity example is carried out to show a case where
twoway methods are unable to provide a unique solution, whereas EPOS gives
correct results.

Grung, B., & Kvalheim, O.M. (1995b).
Rank mapping of threeway multicomponent profiles.
Chemometrics and Ingelligent Laboratory Systems, 29,
223232.
In this work, fast graphic procedures for assessment of number of
analytes (chemical rank) in
local regions of threeway unresolved multicomponent profiles are
presented. Collapsing the
threeway profile along one direction by matrix summation and evolving
rank analysis on the
resulting matrix represents a first step for the approach developed here.
This procedure is
efficient for detecting analytes with low net analytical signal compared
to the noise level.
The effect of collapsing is a reduction of random noise due to
cancellations, while structure
from analytes is enhanced because of the additive effect of many similar
contributions. In
the second step, slicing of the threeway data array into complete two
way arrays (matrices)
is performed. Projections are then used across these matrices to generate
a rank map. The
strategy developed in this work provides a rapid approach to rank
analysis and permits the
analyst to take into account that the analytical information is normally
not uniformly
distributed in the three directions.

Gruvaeus, G., Wainer, H., & Snyder, F. (1971).
TREMOD: A 360/75 FORTRAN program for threemode factor analysis.
Behavioral Science, 16, 421422.
Description of Tucker's threemode factor analysis.

Gui, M., Rutan, S.C., & Agbodjan, A. (1995).
Kinetic detection of overlapped amino acids in
thinlayer chromatography with a direct trilinear
decomposition method.
Analytical Chemistry, 67, 32933299.
Kinetic fluorescence detection (KFD) was employed
to determine the concentrations of two overlapped
components after a thinlayer chromatographic
(TLC) separation. Two amino acids, glycine and
glutamine, were used as model analytes. These
species exhibited very similar retardation factors
(R_{f}) under our experimental conditions. A
reaction that produced a fluorescent product was
performed subsequent to the separation, the
reaction can be described by the following scheme:
Amino acid + OPA > fluorescent product >
nonfluorescent product. Here OPA stands for
ophthalaldehyde. The kinetic profile of the
reaction was monitored with a chargecoupled
device camera by taking sequential images of the
separation medium after the reaction starts. a
direct. Trilinear decomposition (TLD) method was
used to analyze the resulting thirdorder data
that consist of fluorescence intensities as a
function of elution distance, reaction time, and
sample number, this approach was used to determine
the initial concentrations of the two overlapped
components based on the different kinetics and
retention exhibited by these species. This paper
discusses the kinetic approach and the
applicability and limitations of the direct
trilinear decomposition (TLD) method using both
synthetic and experimental data, efforts to
optimize the experimental conditions are also
reported. The major focus of this work is to
explore the application of this novel kinetic
fluorescence detection method for TLC separations
(TLCKFD) combined with the TLD data analysis
method.

Guimet, F., Ferre, J., Boque, R., & Rius, F. X. (2004).
Application of unfold principal component analysis and parallel factor analysis to the
exploratory analysis of olive oils by means of excitationemission matrix fluorscence
spectroscopy.
Analytical Chemistry, 515, 7885.
Discrimination between virgin olive oils and pure olive oils is of primary
importance for controlling adulterations. Here, we show the potential usefulness of two
multiway methods, unfold principal component analysis (UPCA) and parallel factor analysis
(PARAFAC), for the exploratory analysis of the two types of oils. We applied both methods
to the excitationemission fluorescence matrices (EEM) of olive oils and then compared the
results with the ones obtained by multivariate principal component analysis (PCA) based on
a fluorescence spectrum recorded at only one excitation wavelength. For UPCA and PARAFAC,
the ranges studied were lambda(ex) = 300400nm, lambda(em) = 400695 nm and lambda(ex) =
300400 nm, lambda(em) = 400600 nm. The first range contained chlorophylls, whose peak was
much more intense than those of the rest of species. The second range did not contain the
chlorophylls peak but only the fluorescence spectra of the remaining compounds (oxidation
products and Vitamin E). The threecomponent PARAFAC model on the second range was found to
be the most interpretable. With this model, we could distinguish well between the two
groups of oils and we could find the underlying fluorescent spectra of three families of
compounds.

Guiteras, J., Beltran, J. L., & Ferrer, R. (1998).
Quantitative multicomponent analysis of polycyclic aromatic hydrocarbons in water samples.
Analytica Chimica Acta, 361, 233240.
A comparative study between three different multivariate calibration methods,
classical least squares (CLS), principal component regression (PCR) and partial least squares
regression (PLSR) was carried out. The calibration models for all three methods were obtained
from a combination of two synchronous fluorescence spectra (recorded at 50 and 100 nm wavelength
increments) for each standard of a calibration set of 70 standards, each containing ten polycyclic
aromatic hydrocarbons (anthracene, benz[a]anthracene, benzo[a]pyrene, chrysene, fluoranthene,
fluorene, naphthalene, perylene, phenanthrene and pyrene). The predictions of the model were compared
with the relative root mean squared difference (RRMSD) obtained from the results of an external
validation set, formed by 15 independent mixtures. Finally, the PLSR and PCR models were used for
the determination of the above mentioned PAHs in spiked natural water samples at concentration
levels between 4 and 20 ng ml(1). Recoveries ranged from 80 to 120% in most cases, although
fluorene gave significantly lower results.

Gurden, S. P., Lage, E. M., de Faria, C. G., Joekes, I., & Ferreira, M. M. C. (2003).
Analysis of video images from a gasliquid transfer experiment: a comparison of
PCA and PARAFAC for multivariate image analysis.
Journal of Chemometrics, 17, 400412.
The use of chemical imaging is a developing area which has potential benefits for
chemical systems where spatial distribution is important. Examples include processes
in which homogeneity is critical, such as polymerizations, pharmaceutical powder
blending and surface catalysis, and dynamic processes such as the study of diffusion
rates or the transport of environmental pollutants. Whilst single images can be
used to determine chemical distribution patterns at a given point in time, dynamic
processes can be studied using a sequence of images measured at regular time intervals,
i.e. a movie. Multivariate modeling of image data can help to provide insight into
the important chemical factors present. However, many issues of how best to apply
these models remain unclear, especially when the data arrays involved have four or
five different dimensions (height, width, wavelength, time, experiment number, etc.).
In this paper we describe the analysis of video images recorded during an experiment
to investigate the uptake Of CO2 across a free airwater interface. The use of PCA
and PARAFAC for the analysis of both single images and movies is described and some
differences and similarities are highlighted. Some other image transformation techniques,
such as chemical mapping and histograms, are found to be useful both for pretreatment
of the raw data and for dimensionality reduction of the data arrays prior to further
modeling. Copyright (C) 2003 John Wiley Sons, Ltd.

Gurden, S.P., Westerhuis, J.A., Bijlsma S., & Smilde, A.K.
(2001).
Modelling of spectroscopic batch process data using grey models to
incorporate external information. Journal of Chemometrics,
15, 101121.
In both analytical and process chemistry, one common aim is to build models
describing measured data. In cases where additional information about the
chemical system is available, this can be incorporated into the model with the
aim of improving model fit and interpretability. A model which consists of a
'hard' or 'white' part describing known sources of variation and a 'soft' or
'black' part describing unknown sources of variation is called a 'grey' model.
In this paper the use of a grey model is demonstrated using data from a first
order chemical batch reaction monitored by UVvis spectroscopy. The resultant
threeway data matrix is modelled using a Tucker3 structure, and external
information about the spectroscopically active compounds is incorporated in the
form of constraints on the model parameters with additional restrictions on the
Tucker3 core matrix. The grey model is then used to analyse new batches.
Different approaches to building grey models are described and some of their
properties discussed.

Gurden, S.P., Westerhuis, J.A., Bro, R., & Smilde, A.K. (2001).
A comparison of multiway regression and scaling methods.
Chemometrics and Intelligent Laboratory Systems, 59, 121136.
Recent years have seen an increase in the number of regression problems
for which the predictor and/or response arrays have orders higher than two,
i.e. multiway data. Examples are found in, e.g. industrial batch process
analysis, chemical calibration using secondorder instrumentation and
quantitative structureactivity relationships (QSAR). As these types of
problems increase in complexity in terms of both the dimensions and the
underlying structures of the data sets, the number of options with respect
to different types of scaling and regression models also increases. In
this article, three methods for multiway regression are compared: unfold
partial least squares (PLS), multilinear PLS and multiway covariates
regression (MCovR). All three methods differ either in the structural
model imposed on the data or the way the model components are calculated.
Three methods of scaling multiway arrays are also compared, along with
the option of applying no scaling. Three data sets drawn from industrial
processes are used in the comparison. The general conclusion is that the
type of data and scaling used is more important than the type of
regression model used in terms of predictive ability. The models do
differ, however, in terms of interpretability.

Gurden, S.P., Westerhuis, J.A., & Smilde, A.K. (2002).
Monitoring of Batch Processes Using Spectroscopy
American Institute of Chemical Engineers Journal, 48, 22832297.
There is an increasing need for new techniques for the understanding, monitoring
and the control of batch processes. Spectroscopy is now becoming established as
a means of obtaining realtime, highquality chemical information at frequent time
intervals and across a wide range of industrial applications. In this article, the
role of spectroscopy for batch process monitoring is discussed in terms of both
current and potential advances. The emphasis is on how to handle the measured data
to extract maximum information for improved process performance and efficiency.
In particular, the use of spectroscopy for statistical proces monitoring is detailed
and considered as complementary to the use of engineering process data. A case
study of the ultravioletvisible monitoring of a firstorder biochemical conversion
reaction is described, as well as the advantages of spectroscopy for process fault
detection and diagnosis. Future prospects for the use of online spectroscopy are
also discussed.

Guterres, M. V., Volpe, P. O. L. & Ferreira, M. M. C. (2004).
Multiwav calibration for creatinine determination in human serum using the Jaffe reaction
Applied Spectroscopy, 58, 5460.
Secondorder calibration and multivariate spectroscopickinetic measurements in
the visible region are proposed to improve the Jaffe method for creatinine assay.
Analyses performed on synthetic mixtures containing bilirubin, glucose, and albumin
confirm that secondorder calibration is useful for creatinine determination in
human serum. Quantitative determinations of creatinine with the parallel factor
analysis (PARAFAC) and direct trilinear decomposition (TLD) methods were compared.
It is shown that both methods can be used for creatinine determination in human
serum, with an SEP (squared error of prediction) of 2.22 and coefficient of variability
of 6.14% for PARAFAC, and an SEP of 2.38 and coefficient of variability of 6.57 %
for TLD.

Gutmanas, A., Jarvoll, P., Orekhov, V. Y., & Billeter, M. (2002).
Threeway decomposition of a complete 3D N15NOESYHSQC.
Journal of Biomolecular NMR, 24, 191201.
Threeway decomposition is applied for the structural analysis of a
complete threedimensional N15NOESYHSQC of the 128 residues long protein azurin. The
procedure presented includes decomposition using the software MUNIN, providing an initial
characterization of the complete spectrum by 355 components. This is followed by
postprocessing yielding a final list of 149 components, 123 of which characterize 1859
NOE peaks from backbone NH groups. Components from threeway decomposition are defined
as direct products of onedimensional shapes along the three dimensions. Thus, a complete
set of distance constraints from this spectrum is obtained by onedimensional peak
picking of the shapes along the NOE dimension. Correctness and completeness of this set
of NOEs are tested for all backbone amide groups against both an independent peak picking
algorithm and the threedimensional crystal structure of azurin, and a coincidence of
about 95% is observed. Automated 'demixing' of components that are 'mixed' in a complex
manner due to overlap of the HN and/or N15 frequencies is illustrated.
Go to other sections of the Abstracts
 A  B 
C  D 
E  F 
G  H 
I  J 
K  L 
M  N 
O  P 
Q  R 
S  T 
U  V 
W  X 
Y  Z 
Top 
 Algemene
Pedagogiek 
Centre for Child and
Family Studies 
Department of Educational
Sciences 
The ThreeMode Company 
ThreeMode bibliography

P.M. Kroonenberg
Education and Child Studies, Leiden University
Wassenaarseweg 52, 2333 AK Leiden, The Netherlands
Tel. *31715273446/5273434 (secr.); fax *31715273945
Email:
kroonenb@fsw.leidenuniv.nl
First version : 12/02/1997;