, and after it the principal component analysis (PCA) method ( Wold et al., 1987 ; Martens and Næs, 1991 ), the clusteranalysis (CA) method ( Heise and Winzen, 2002 ), and the polar qualification system (PQS) method ( Kaffka and Seregély, 2002 ) were
This paper describes experimental results through multivariate statistical methods that might reveal outliers that are rarely
taken into account by analysts. The results were submitted to three procedures to detect outliers: Mahalanobis distance, MD,
cluster analysis, CA, and principal component analysis, PCA. The results showed that although CA is one of the procedures
most often used to identify outliers, it can fail by not showing the samples that are easily identified as outliers by other
methods, like MD. Mahalanobis distance proved to be the simpler application, with sensitive procedures to identify outliers
in multivariate datasets.
Chemical and mineralogical analysis was performed on ceramics and clay samples from Barracão archaeological site located in
Baixo São Francisco River by means of instrumental neutron activation analysis (INAA) and by differential scanning calorimetry
(DSC). The data set was studied by means of cluster analysis (CA) and discriminant analysis (DA). The results showed that
the raw material used in ceramics is not local. By using DSC it was possible to discover that the principal minerals in the
samples are quartz, feldspars, mica and kaolinite.
In 1999 investigation, the needles of Sabina chinesis L. from 11 hot spot tourist cities extending over China were collected as biomonitors for evaluation of atmospheric elemental differences in those cities. Thirty five elements were determined by instrumental neutron activation analysis (INAA). The experimental data reflected the environmental alteration in the cities that have different geographical position, population burden and industrialized level. Eventually by using the data processed by factor analysis (FA) and subsequent cluster analysis (CA), the cities were grouped into six sorts of the urban atmospheric deposition with corresponding elements.
Near-infrared (NIR) spectrophotometers with different optical arrangements were used to measure the diffusely reflected electromagnetic radiation of different types of food additives in polyethylene (PE) foils. Eight compounds, frequently used in food process – but also in other industries –, were measured by different NIR spectrophotometers. The detected (‘as is’) and mathematically transformed (by scatter correction, second derivative combined with smoothing) NIR spectra were processed with multivariate data analysis (MDA). In this matter, unsupervised methods like principal component analysis (PCA) and cluster analysis (CA) were used, which techniques do not require prior information and reference measurements. The aim of the present study was to distinguish food additives by the help of the applied chemometric methods based on NIR spectra detected via the PE foils. Results indicated that distinction of different food additives and compounds with NIR methods is possible not only with the conventional sample preparations and handlings, but also without breaking the packaging.
The study deals with the application of cluster analysis (CA) and non-parametric tests (Shapiro–Wilk, Kruskal–Wallis, Dunn,
U Mann–Whitney) to classify and interpret of a monitoring data set for Odra River water quality assessment based on concentration
values of radiochemical parameters. The data set represents results for 3 alpha emitters (210Po, 238U and 239+240Pu) measured in surface water samples collected at 13 different sampling locations (5 in major Odra stream while 8 in Odra
tributaries) within four seasons: winter, spring, summer and autumn, in the framework of 1 year-term quality monitoring research.
The correlation analysis of polonium, uranium and plutonium data indicates that significant values of Spearman’s correlation
coefficient appears between 210Po and 239+240Pu (r = 0.55 in autumn and 0.77 in winter as well as 0.49 in all year), while statistical significant correlation between uranium
and plutonium as well as uranium and polonium were not found. In the Odra drainage basin, the biggest differences were observed
in the case of 238U. The hypothesis about possible geographic and seasonal differences between concentration of 210Po, 238U and 239+240Pu in the Odra River catchment area was verified by cluster analysis (CA). Finally, to asses if there are statistically significant
differences in mean concentration value of 210Po, 238U and 239+240Pu for Vistula and Odra Rivers drainage basins were obtained by used of the non-parametric tests. Comparing to Vistula catchment
area, statistically different concentration of 210Po and 239+240Pu in all year was observed for river samples collected on Odra drainage basin.
Meriç River Delta is located in the Thrace Region of Turkey, and it is one of the most important wetlands worldwide. Gala and Sığırcı Lakes, which are known as significant lakes in Turkey in terms of especial biodiversity, are located in the Meriç River Delta and they are the main lentic factors of the system. The aim of this study was to evaluate the essential and toxic element bioaccumulation levels in fishes of Gala and Sığırcı Lakes from a statistical perspective by investigating a total of 25 macro- and micro-element bioaccumulations. One-Way ANOVA Test (OWAT) was applied to detected data in order to determine the statistical differences of element bioaccumulations among the fish species. Cluster analysis (CA) was also applied to detected data in order to classify the investigated elements in terms of bioaccumulation levels in fish tissues. According to the results of OWAT, although statistical differences were not recorded among the fish species in terms of essential element levels, significant statistical differences were recorded in terms of toxic element levels (P<0.05). According to the results of CA, 5 statistically significant clusters were formed, which were named as “Most intense elements”, “Second most intense elements”, “Moderate intense elements”, “Second rarest elements”, and “Rarest elements”. It was also found that toxic element bioaccumulation rates in fishes of Gala Lake were significantly higher than in fishes of Sığırcı Lake (P<0.05).
Multivariate methods were successfully employed in a comprehensive scientometric analysis of geostatistics research, and the
publications data for this research came from the Science Citation Index and spanned the period from 1967 to 2005. Hierarchical
cluster analysis (CA) was used in publication patterns based on different types of variables. A backward discriminant analysis
(DA) with appropriate statistical tests was then conducted to confirm CA results and evaluate the variations of various patterns.
For authorship pattern, the 50 most productive authors were classified by CA into 4 groups representing different levels,
and DA produced 92.0% correct assignment with high reliability. The discriminant parameters were mean impact factor (MIF),
annual citations per publication (ACPP), and the number of publications by the first author, for country/region pattern, CA
divided the top 50 most productive countries/regions into 4 groups with 95.9% correct assignments, and the discriminant parameters
were MIF, ACCP, and independent publication (IP); for institute pattern, 3 groups were identified from the top 50 most productive
institutes with nearly 88.0% correct assignment, and the discriminant parameters were MIF, ACCP, IP, and international collaborative
publication; last, for journal pattern, the top 50 most productive journals were classified into 3 groups with nearly 98.0%
correct assignment, and its discriminant parameters were total citations, impact factor and ACCP. Moreover, we also analyzed
general patterns for publication document type, language, subject category, and publication growth.
Activity concentrations using gamma-ray spectrometer and distributions of natural radionuclides in soil samples collected
were investigated to assess the environmental radioactivity and characterization of radiological hazard. The average concentrations
of 238U, 232Th series and 40K in the 5 cm depth soil were 22.53, 33.43 and 406.62 Bq kg−1, respectively, which was within world median ranges in the UNSCEAR 2000 report. The average absorbed dose rate estimated
by soil activity and annual effective doses were 49.32 nGy h−1 and 60.48 μSv, respectively. Since the soil is an important building material, the mean radium equivalent activity (Raeq), external (Hex) and internal (Hin) hazard index using various models given in the literature for the study area were evaluated as 101.72 Bq kg−1, 0.27 and 0.34, respectively, which were below the recommended limits. The effects of pH value, conductivity, true density
and textural properties of soil samples on the natural radionuclide levels were also studied. The application of cluster analysis
(CA) and principal component analysis (PCA), coupled with Pearson correlation coefficient analysis, were utilized to analyze
the data, identify and clarify the effects of physico-chemical properties on natural radioactivity levels. The CA and PCA
results showed that the former method yielded three distinctive groups of the soil variables whereas the latter one yielded
the number of variables into three factors with 87.5% variance explanation.
origin 3.4 E-nose data of PEL The CA, LDA, and PCA analysis results of the E-nose were as follows. Based on the significant e-nose response of W5S, W1W, and W2W to PEL ( Fig. 2A ), clusteranalysis (CA) was utilised to describe the variations in the aroma