Comparing Different Approaches for Clustering Categorical Data

There are different ways to do cluster analysis of categorical data in the literature and the choice among them is strongly related to the aim of the researcher, if we do not take into account time and economical constraints. Main approaches for clustering are usually distinguished into model-based and distance-based methods: the former assume that objects belonging to the same class are similar in the sense that their observed values come from the same probability distribution, whose parameters are unknown and need to be estimated; the latter evaluate distances among objects by a defined dissimilarity measure and, basing on it, allocate units to the closest group. In clustering, one may be interested in the classification of similar objects into groups, and one may be interested in finding observations that come from the same true homogeneous distribution. But do both of these aims lead to the same clustering? And how good are clustering methods designed to fulfil one of these aims in terms of the other? In order to answer, two approaches, namely a latent class model (mixture of multinomial distributions) and a partition around medoids one, are evaluated and compared by Adjusted Rand Index, Average Silhouette Width and Pearson-Gamma indexes in a fairly wide simulation study. Simulation outcomes are plotted in bi-dimensional graphs via Multidimensional Scaling; size of points is proportional to the number of points that overlap and different colours are used according to the cluster membership.

Software Cartography: Thematic Software Visualization with Consistent Layout

Software visualizations can provide a concise overview of a complex software system. Unfortunately, as software has no physical shape, there is no `natural' mapping of software to a two-dimensional space. As a consequence most visualizations tend to use a layout in which position and distance have no meaning, and consequently layout typically diverges from one visualization to another. We propose an approach to consistent layout for software visualization, called Software Cartography, in which the position of a software artifact reflects its vocabulary, and distance corresponds to similarity of vocabulary. We use Latent Semantic Indexing (LSI) to map software artifacts to a vector space, and then use Multidimensional Scaling (MDS) to map this vector space down to two dimensions. The resulting consistent layout allows us to develop a variety of thematic software maps that express very different aspects of software while making it easy to compare them. The approach is especially suitable for comparing views of evolving software, as the vocabulary of software artifacts tends to be stable over time. We present a prototype implementation of Software Cartography, and illustrate its use with practical examples from numerous open-source case studies.

Metacommunity Dynamics Of Benthic Macroinvertebrates In A Large River

Metacommunity ecology focuses on the interaction between local communities and is inherently linked to dispersal as a result. Within this framework, communities are structured by a combination of in-site responses to the immediate environment (species sorting), stochasticity (patch dynamics), and connections to other communities via distance between communities and dispersal (neutrality), and source-sink dynamics (mass effects; see Chapter 1 for a detailed description of metacommunity theory, the study site, and macroinvertebrate communities found). In Chapter 2 I describe spatial scale of study and dispersal ability as both have the ability to influence the degree to which communities interact. However, little is known about how these factors influence the importance of all metacommunity dynamics. I compared dispersal mode of immature aquatic insects and dispersal ability of winged adults across multiple spatial scales in a large river. The strongest drivers of river communities were patch dynamics, followed by species sorting, then neutrality. Active dispersers during aquatic lifestages on average exhibited lower patch dynamics, higher species sorting, and significant mass effects compared to passive dispersers. Active and strong dispersers also had a scale-independent influence of neutrality, while neutrality was stronger at broader spatial scale for passive and weak dispersers. These results indicate as dispersal ability increases patch dynamics decreases, species sorting increases, and neutrality should decrease. The perceived influence of neutrality may also be dependent on spatial scale and dispersal ability. In Chapter 3 I describe how river benthic macroinvertebrate communities may influence tributary invertebrate communities via adult flight and tributaries may influence mainstem communities via immature drift. This relationship may also depend on relative mainstem and tributary size, as well as abiotic tributary influence on mainstem habitat. To investigate the interaction between a larger river and tributary I sampled mainstem benthic invertebrate communities and quantified habitat of a 7th order river (West Branch Susquehanna River) above and below a 5th order tributary confluence, as well as 0.95-3.2 km upstream in the tributary. Non-metric multidimensional scaling showed similar patterns of clustering between sampling locations for both habitat characteristics and invertebrate communities. In addition, mainstem river communities and habitat directly downstream of the tributary confluence cluster tightly together, intermediate between tributary and mid-channel river samples. In Bray-Curtis dissimilarity comparisons between tributary and mainstem river communities the furthest upstream tributary communities were least similar to river communities. Middle tributary samples were also closest by Euclidean distance to the upstream mainstem riffle and exhibited higher similarity to mid-channel samples than the furthest downstream tributary communities. My results indicate river and tributary benthic invertebrate communities may interact and likely result in direct and indirect mass effects of a tributary on the downstream mainstem community by invertebrate drift and habitat restructuring via material delivery from the tributary. I also showed likely direct effects of adult dispersal from the river and oviposition in proximal tributary locations where Euclidian, rather than river, distance may be more important in determining river-tributary interactions.

Tributary and Mainstem Benthic Macroinvertebrate Communities Linked By Direct Dispersal and Indirect Habitat Alteration

Benthic communities in tributary-mainstem networks might interact via downstream drift of invertebrates or material from tributaries and adult dispersal from the mainstem. Depending on the strength of these interactions, mainstem downstream communities are expected to be more similar to tributary communities due to drift or habitat alteration. Communities not connected by flow are expected to be similar due to adult dispersal but decreasing in similarity with distance from the mainstem. We investigated interactions between invertebrate communities of a 7th order river and 5th order tributary by comparing benthic community structure in the river upstream and downstream of the tributary confluence and upstream in the tributary. Non-metric multidimensional scaling showed invertebrate communities and habitat traits from river locations directly downstream of the tributary clustered tightly, intermediate between tributary and mid-channel river locations. In addition, Bray-Curtis dissimilarity increased between the mainstem and tributary with distance upstream in the tributary. Our results indicate that similarities between mainstem and tributary communities are potentially caused by direct mass effects from tributary to downstream mainstem communities by invertebrate drift and indirect mass effects by habitat restructuring via material delivery from the tributary, as well as potential effects of adult dispersal from the river on proximal tributary communities.

Patterned-ground plant communities along a bioclimate gradient in the High Arctic, Canada

Non-sorted circles, non-sorted polygons, and earth hummocks are common ground-surface features ill arctic regions. The), are caused by a variety of physical processes that Occur in permafrost regions including contraction cracking and frost heave. Here we describe the vegetation of patterned-ground forms on zonal sites at three location!: along an N-S transect through the High Arctic of Canada. We made 75 releves on patterned-ground features (circles, polygons, earth hummocks) and adjacent tundra (Interpolygon, intercircle, interhummock areas) and identified and classified the vegetation according to the Braun-Blanquet Method. Environmental factors were correlated with the vegetation data using a nonmetric multidimensional scaling ordination (NMDS). We identified eleven commnunities: (1) Puccinellia angustata-Papaver radicalum community in xeromesic non-sorted polygons of subzone A of the Circumpolar Arctic Vegetation Map; (2) Saxifraga-Parmelia omphalodes ssp. glacialis community in hydromesic interpolygon areas of subzone A; (3) Hypogymnia subobscura-Lecanora epibryon community In xeromesic non-sorted polygons of subzone B; (4) Orthotrichum speciosum-Salix arctica community In xeromesic interpolygon areas of subzone B; (5) Cochlearia groenlandica-Luzula nivalis community in hydromesic earth Mocks Of subzone B; (6) Salix arctica-Eriophorum angustifolium ssp. triste community in hygric earth hummocks of subzone 13; (7) Puccinellia angustata-Potentilla vahliana community in xeromesic non-sorted circles and bare patches of subzone Q (8) Dryas integrifolia-Carex rupestris community in xeromesic intercircle areas and vegetated patches of subzone C; (9) Braya glabella ssp. purpurascens-Dryas integrifolia community In hydromesic non-sorted circles of subzone Q (10) Dryas integrifolia-Carex aquatilis community in hydromesic intercircle areas of subzone C; and (11) Eriophorum angustifolium ssp. triste-Carex aquatilis community ill hygric intercircle areas of subzone C. The NMDS ordination displayed the vegetation types with respect to complex environmental gradients. The first axis of the ordination corresponds to a complex soil moisture gradient and the second axis corresponds to a complex geology/elevation/climate gradient. The tundra plots have a greater moss and graminoid cover than the adjacent frost-heave communities. In general, frost-heave features have greater thaw depths, more bare ground, thinner organic horizons, and lower soil moisture than the surrounding tundra. The morphology of the investigated patterned ground forms changes along the climatic gradient, with non-sorted pollygons dominating in the northernmost sites and non-sorted circles dominating, in the southern sites.

Consistent Layout for Thematic Software Maps

Software visualizations can provide a concise overview of a complex software system. Unfortunately, since software has no physical shape, there is no “natural“ mapping of software to a two-dimensional space. As a consequence most visualizations tend to use a layout in which position and distance have no meaning, and consequently layout typical diverges from one visualization to another. We propose a consistent layout for software maps in which the position of a software artifact reflects its \emph{vocabulary}, and distance corresponds to similarity of vocabulary. We use Latent Semantic Indexing (LSI) to map software artifacts to a vector space, and then use Multidimensional Scaling (MDS) to map this vector space down to two dimensions. The resulting consistent layout allows us to develop a variety of thematic software maps that express very different aspects of software while making it easy to compare them. The approach is especially suitable for comparing views of evolving software, since the vocabulary of software artifacts tends to be stable over time.

A synchronic lexical study of the Ede language continuum of West Africa: The effects of different similarity judgment criteria

In the context of a synchronic lexical study of the Ede varieties of West Africa, this paper investigates whether the use of different criteria sets to judge the similarity of lexical features in different language varieties yields matching conclusions regarding the relative relationships and clustering of the investigated varieties and thus leads to similar recommendations for further sociolinguistic research. Word lists elicited in 28 Ede varieties were analyzed with the inspection method. To explore the effects of different similarity judgment criteria, two different similarity judgment criteria sets were applied to the elicited data to identify similar lexical items. The quantification of these similarity decisions led to the computation of two similarity matrices which were subsequently analyzed by means of correlation analysis and multidimensional scaling. The findings of this analysis suggest compatible conclusions regarding the relative relationships and clustering of the investigated Ede varieties. However, the matching clustering results do not necessarily lead to the same recommendations for more in-depth sociolinguistic research, when interpreted in terms of an absolute lexical similarity threshold. The indicated ambiguities suggest the usefulness of focusing on the relative, rather than absolute in establishing recommendations for further sociolinguistic research.

Investigating Research Streams of Conjoint Analysis: A Bibliometric Study

Conjoint analysis (CA) is one of the most important methods for preference elicitation. In this paper we investigate the intellectual structure within the conjoint analytical research community. Analyses based on single papers provide a method-based overview of streams of conjoint research. By using novel bibliometric techniques in this field we complement findings of existing reviews. We use co-citation and factor analysis of the most cited articles in SSCI to identify the most important articles and research streams. Seven research streams are revealed which are visualized by means of multidimensional scaling. Tables and graphics reveal the disciplinary affiliations of contributors to CA, the special structure within the classes as well as links between them.

DNA-based identification of novel bovine casein gene variants

In cattle, at least 39 variants of the 4 casein proteins (α(S1)-, β-, α(S2)- and κ-casein) have been described to date. Many of these variants are known to affect milk-production traits, cheese-processing properties, and the nutritive value of milk. They also provide valuable information for phylogenetic studies. So far, the majority of studies exploring the genetic variability of bovine caseins considered European taurine cattle breeds and were carried out at the protein level by electrophoretic techniques. This only allows the identification of variants that, due to amino acid exchanges, differ in their electric charge, molecular weight, or isoelectric point. In this study, the open reading frames of the casein genes CSN1S1, CSN2, CSN1S2, and CSN3 of 356 animals belonging to 14 taurine and 3 indicine cattle breeds were sequenced. With this approach, we identified 23 alleles, including 5 new DNA sequence variants, with a predicted effect on the protein sequence. The new variants were only found in indicine breeds and in one local Iranian breed, which has been phenotypically classified as a taurine breed. A multidimensional scaling approach based on available SNP chip data, however, revealed an admixture of taurine and indicine populations in this breed as well as in the local Iranian breed Golpayegani. Specific indicine casein alleles were also identified in a few European taurine breeds, indicating the introgression of indicine breeds into these populations. This study shows the existence of substantial undiscovered genetic variability of bovine casein loci, especially in indicine cattle breeds. The identification of new variants is a valuable tool for phylogenetic studies and investigations into the evolution of the milk protein genes.

A comparative analysis of CHIP Perinatal policy in twelve states

The purpose of this comparative analysis of CHIP Perinatal policy (42 CFR § 457) was to provide a basis for understanding the variation in policy outputs across the twelve states that, as of June 2007, implemented the Unborn Child rule. This Department of Health and Human Services regulation expanded in 2002 the definition of “child” to include the period from conception to birth, allowing states to consider an unborn child a “targeted low-income child” and therefore eligible for SCHIP coverage. ^ Specific study aims were to (1) describe typologically the structural and contextual features of the twelve states that adopted a CHIP Perinatal policy; (2) describe and differentiate among the various designs of CHIP Perinatal policy implemented in the states; and (3) develop a conceptual model that links the structural and contextual features of the adopting states to differences in the forms the policy assumed, once it was implemented. ^ Secondary data were collected from publicly available information sources to describe characteristics of states’ political system, health system, economic system, sociodemographic context and implemented policy attributes. I posited that socio-demographic differences, political system differences and health system differences would directly account for the observed differences in policy output among the states. ^ Exploratory data analysis techniques, which included median polishing and multidimensional scaling, were employed to identify compelling patterns in the data. Scaled results across model components showed that economic system was most closely related to policy output, followed by health system. Political system and socio-demographic characteristics were shown to be weakly associated with policy output. Goodness-of-fit measures for MDS solutions implemented across states and model components, in one- and two-dimensions, were very good. ^ This comparative policy analysis of twelve states that adopted and implemented HHS Regulation 42 C.F.R. § 457 contributes to existing knowledge in three areas: CHIP Perinatal policy, public health policy and policy sciences. First, the framework allows for the identification of CHIP Perinatal program design possibilities and provides a basis for future studies that evaluate policy impact or performance. Second, studies of policy determinants are not well represented in the health policy literature. Thus, this study contributes to the development of the literature in public health policy. Finally, the conceptual framework for policy determinants developed in this study suggests new ways for policy makers and practitioners to frame policy arguments, encouraging policy change or reform. ^

Counts of benthos communities in Kongsfjorden in 2009

In order to reveal the structure of the sparsely known deeper sublittoral hard bottom communities of glacial Kongsfjorden, the macroepibenthos from six depth zones (30-200 m) was analysed. A total of 180 still images derived from 6-h video recorded at the Kongsfjordneset remotely operated vehicle station were assessed quantitatively. Overall 27 mainly suspension-feeding species/taxa were observed. Of these, two-thirds have an arcto-boreal distribution, while the remainder are cosmopolitan. The overall mean epibenthos abundance was 33 ind./m**2 with maximum values at 150 m depth (97.9 ind./m**2). The majority of the taxa inhabited the entire depth range. Encrusting red algae, an unidentified sponge and the sea anemone Urticina eques, characterized the assemblage of the shallow zone. The sea anemones Hormathia spp. were important below 30 m, the Serpulid polychaete Protula tubularia was characteristic for the community below 50 m and the demosponge Haliclona sp. was a key taxon between 100 and 200 m depth. Cluster analysis and non-metrical multidimensional scaling based on abundance data showed differences between the assemblages along the bathymetric gradient, but only in the shallower depths in relation to the substratum surface incline. As surface and tidal current impacts attenuate with increasing depth, there is a gradual trend from robust key species towards more fragile ones (i.e. P. tubularia), in line with the 'Physical control hypothesis'.

Environmental variables and multivariate seabed classification maps via k means clustering of Potter Cove, Antarctica, link to input and result files in GeoTIFF format

This study subdivides the Potter Cove, King George Island, Antarctica, into seafloor regions using multivariate statistical methods. These regions are categories used for comparing, contrasting and quantifying biogeochemical processes and biodiversity between ocean regions geographically but also regions under development within the scope of global change. The division obtained is characterized by the dominating components and interpreted in terms of ruling environmental conditions. The analysis includes in total 42 different environmental variables, interpolated based on samples taken during Australian summer seasons 2010/2011 and 2011/2012. The statistical errors of several interpolation methods (e.g. IDW, Indicator, Ordinary and Co-Kriging) with changing settings have been compared and the most reasonable method has been applied. The multivariate mathematical procedures used are regionalized classification via k means cluster analysis, canonical-correlation analysis and multidimensional scaling. Canonical-correlation analysis identifies the influencing factors in the different parts of the cove. Several methods for the identification of the optimum number of clusters have been tested and 4, 7, 10 as well as 12 were identified as reasonable numbers for clustering the Potter Cove. Especially the results of 10 and 12 clusters identify marine-influenced regions which can be clearly separated from those determined by the geological catchment area and the ones dominated by river discharge.

Benthic isopod species in the Ross Sea and their distribution in the Southern Ocean

In this study isopod species of the Ross Sea were investigated. Literature until May 2008 was checked to provide an overview of all known and described species in the Ross Sea. This species checklist was then enlarged through material of the 19th Italica expedition in 2004. During this expedition for the first time a small mesh net (500 µm) was used. Nine thousand four hundred and eighty one isopod specimens were collected during this expedition. Through this material the number of isopod species in the Ross Sea increased from 42 to 117 species, which belong to 20 families and 49 genera. Fifty-six percentage of the isopods species collected during the Italica expedition are new to science. The zoogeography of the 117 species was investigated. A non-transformed binary presence-absence data matrix was constructed using the Bray-Curtis coefficient. The results were displayed in a cluster analysis and by nonmetric multidimensional scaling (MDS). This paper gives a first insight into the occurrence and distribution of the isopod species of the Ross Sea.

Sublittoral and bathyal Harpacticoida (Crustacea: Copepoda) of the Magellan region

Two expeditions, undertaken in 1994 and 1996, provided quantitatively sampled material of sublittoral and bathyal meiobenthos from the Paso Ancho of the Straits of Magellan, the Beagle Channel, and the Patagonian continental slope (Chile). To investigate whether these distinct geographic areas might also be characterised by different harpacticoid assemblages, qualitative and quantitative analyses of Copepoda Harpacticoida were carried out. At supraspecific level 25 harpacticoid families were found, as well as several species that could not yet be assigned to any major harpacticoid taxon. Due to the high amount of collected Harpacticoida, detailed investigations at species level had to be restricted to six taxa, namely the Ancorabolidae, Argestidae, Cletodidae, Diosaccinae, Paramesochridae, and Paranannopinae. The corresponding specimens were assigned to 122 species in 52 genera. More than 80% of them are new to science. Qualitative comparisons of both species composition and species distribution allow the three areas to be distinguished in terms of species richness. However, statistical analyses confirm these results only partly. Similarity analyses applying non-metrical multidimensional scaling, as well as diversity analyses using the rarefaction method, suggest that the observed differences in distribution and diversity patterns are due to small-scale, local conditions, which may overlay possible large-scale ones.

Soft-bottom macrobenthic faunal associations in the southern Chilean glacial fjord complex

Macrobenthic associations were investigated at 29 sampling stations with a semi-quantitative Agassiz trawl, ranging from the South Patagonian Icefield to the Straits of Magellan in the South Chilean fjord system. A total of 1,895 individuals belonging to 131 species were collected. 19 species belong to colonial organisms, mainly Bryozoa (17 species) and Octocorallia (2 species). The phylum Echinodermata was the most diverse in species number (47 species), with asteroids (25 species) and ophiuroids (13 species) being the best represented within this taxon. Polychaeta was the second dominant group in terms of species richness (46 species). Multidimensional scaling ordination (MDS) separated two station groups, one related to fjords and channels off the South Patagonian Icefield and the second one to stations surrounding the Straits of Magellan. 45 species account for 90% of the dissimilarity between these two groups. These differences can mainly be explained by the influence of local environmental conditions determined by processes closely related to the pres- ence/absence of glaciers. Abiotic parameters such as water depth, type of sediment and chemical features of the superficial sediment were not correlated with the numbers of individuals caught by the Agassiz trawl in each group of sampling stations.