Orientador: Alan Rodrigo Panosso / Resumo: A análise de Escritórios de Desenvolvimento Rural (EDR) permite que se tenha uma melhor alocação dos recursos disponíveis, bem como um aumento da eficiência técnica da produção agropecuária, desde a geração de empregos, a produção de matérias-primas para a indústria e o mercado. O objetivo do trabalho foi analisar a eficiência agropecuária de treze municípios (DMU’s - Decision Making Units) pertencentes ao EDR de Andradina, Estado de São Paulo. A análise exploratória dos dados foi realizada, inicialmente, por meio da estatística multivariada, pelas técnicas de análise de agrupamento não hierárquica e pela análise dos componentes principais. Sequencialmente, foi realizada a análise envoltória de dados (Data Envelopment Analysis - DEA), com modelo variável de escala (BCC) e orientação output. Foram utilizados três inputs (terra, trabalho, capital) e um output (produção) totalizando quatro variáveis. A análise multivariada permitiu a formação de dois grupos, em que as variáveis terra, trabalho e produção contribuíram para a separação do agrupamento principal, tendo o município de Valparaíso como destaque. A análise DEA aponta que a maioria das DMU’s trabalhou de maneira ineficiente, tendo apenas 30,77% da amostra eficiente no modelo escolhido (BCC), indicando diferentes níveis tecnológicos das unidades agropecuárias analisadas. / Mestre

No processo de desenvolvimento de novos produtos e serviços, o entendimento de quais demandas são exigidas pelo mercado conduz ao desenvolvimento de projetos com melhores soluções aos clientes. Na busca deste entendimento, técnicas estatísticas multivariadas são utilizadas como suporte para identificar e valorar os requisitos derivados destas demandas. Nesse contexto, o objetivo deste trabalho é apresentar uma abordagem para aplicação de técnicas estatísticas multivariadas no processo de desenvolvimento de novos produtos (PDP). Estas técnicas podem auxiliar as empresas no gerenciamento de requisitos, contribuindo para: (i) coletar e organizar os requisitos do produto; (ii) identificar os requisitos considerados como mais relevantes; (iii) identificar os segmentos de mercado baseado nas características valoradas pelo público-alvo; (iv) verificar associações entre requisitos de um produto e determinadas características do público-alvo. Este trabalho apresenta um exemplo de aplicação contemplando o uso combinado de técnicas estatísticas tais como o método CHAID (Chi-squared Automatic Interaction Detector), análise fatorial, análise conjunta de atributos e análise de correspondência. A demonstração do emprego destas técnicas é realizada no desenvolvimento de um novo produto de limpeza doméstica produzido com características de sustentabilidade. / The deep understanding of market’s requirements, during the manufacturing of new products and/or services, leads to the creation of products of better configuration attend customers’ necessities. Multivariable analysis techniques can be employed to help identifying such consumer preferences. Therefore, the aim of this study is to illustrate an approach to the employment of multivariate statistical procedures on the development of new products (DNP). These techniques can assist companies in managing products’ requirements by helping them to: (i) assemble and categorize products’ requirements; (ii) identify those requirements considered more relevant among all; (iii) identify market sectors based on the aspects most valuable to consumers; (iv) check on associations between one given product and certain features of general customers. The present work illustrates the combined use of statistical techniques such as the CHAID (Chi-squared Automatic Interaction Detector), factorial analysis, conjoint analysis and correspondence analysis. The successful application of these techniques is exemplified with the development of a new domestic cleaning environmental-friendly product.

Cette thèse contribue au domaine de la morphologie mathématique et illustre comment la statistique multivariée et les techniques d'apprentissage numérique peuvent être exploitées pour concevoir un ordre dans l'espace des vecteurs et pour inclure les résultats d'opérateurs morphologiques au processus d'analyse d'images multivariées. En particulier, nous utilisons l'apprentissage supervisé, les projections aléatoires, les représentations tensorielles et les transformations conditionnelles pour concevoir de nouveaux types d'ordres multivariés et de nouveaux filtres morphologiques pour les images multi/hyperspectrales. Nos contributions clés incluent les points suivants :• Exploration et analyse d'ordre supervisé, basé sur les méthodes à noyaux.• Proposition d'un ordre nonsupervisé, basé sur la fonction de profondeur statistique calculée par projections aléatoires. Nous commençons par explorer les propriétés nécessaires à une image pour assurer que l'ordre ainsi que les opérateurs morphologiques associés, puissent être interprétés de manière similaire au cas d'images en niveaux de gris. Cela nous amènera à la notion de décomposition en arrière plan. De plus, les propriétés d'invariance sont analysées et la convergence théorique est démontrée.• Analyse de l'ordre supervisé dans les problèmes de correspondance morphologique de patrons, qui correspond à l'extension de l'opérateur tout-ou-rien aux images multivariées grâce à l‘utilisation de l'ordre supervisé.• Discussion sur différentes stratégies pour la décomposition morphologique d'images. Notamment, la décomposition morphologique additive est introduite comme alternative pour l'analyse d'images de télédétection, en particulier pour les tâches de réduction de dimension et de classification supervisée d'images hyperspectrales de télédétection.• Proposition d'un cadre unifié basé sur des opérateurs morphologiques, pour l'amélioration de contraste et pour le filtrage du bruit poivre-et-sel.• Introduction d'un nouveau cadre de modèles Booléens multivariés en utilisant une formulation en treillis complets. Cette contribution théorique est utile pour la caractérisation et la simulation de textures multivariées. / This thesis contributes to the field of mathematical morphology and illustrates how multivariate statistics and machine learning techniques can be exploited to design vector ordering and to include results of morphological operators in the pipeline of multivariate image analysis. In particular, we make use of supervised learning, random projections, tensor representations and conditional transformations to design new kinds of multivariate ordering, and morphological filters for color and multi/hyperspectral images. Our key contributions include the following points:• Exploration and analysis of supervised ordering based on kernel methods.• Proposition of an unsupervised ordering based on statistical depth function computed by random projections. We begin by exploring the properties that an image requires to ensure that the ordering and the associated morphological operators can be interpreted in a similar way than in the case of grey scale images. This will lead us to the notion of background/foreground decomposition. Additionally, invariance properties are analyzed and theoretical convergence is showed.• Analysis of supervised ordering in morphological template matching problems, which corresponds to the extension of hit-or-miss operator to multivariate image by using supervised ordering.• Discussion of various strategies for morphological image decomposition, specifically, the additive morphological decomposition is introduced as an alternative for the analysis of remote sensing multivariate images, in particular for the task of dimensionality reduction and supervised classification of hyperspectral remote sensing images.• Proposition of an unified framework based on morphological operators for contrast enhancement and salt- and-pepper denoising.• Introduces a new framework of multivariate Boolean models using a complete lattice formulation. This theoretical contribution is useful for characterizing and simulation of multivariate textures.

The work in this dissertation represents an attempt to investigate multiple temporal and spatial scales of inquiry relating to the variability of marine resources throughout the Gulf of Mexico large marine ecosystem (Gulf LME). This effort was undertaken over two spatial extents within the greater Gulf LME using two different time-series of fisheries monitoring data. Case studies demonstrating simple frameworks and best practices are presented with the aim of aiding researchers seeking to reduce errors and biases in scientific decision making. Two of the studies focused on three years of groundfish survey data collected across the West Florida Shelf (WFS), an ecosystem that occupies the eastern portion of the Gulf LME and which spans the entire latitudinal extent of the state of Florida. A third study was related to the entire area covered by the Gulf LME, and explored a 30-year dataset containing over 100 long-term monitoring time-series of indicators representing (1) fisheries resource status and structure, (2) human use patterns and resource extractions, and (3) large- and small-scale environmental and climatological characteristics. Finally, a fourth project involved testing the reliability of a popular new clustering algorithm in ecology using data simulation techniques. The work in Chapter Two, focused on the WFS, describes a quantitatively defensible technique to define daytime and nighttime groundfish assemblages, based on the nautical twilight starting and ending times at a sampling station. It also describes the differences between these two unique diel communities, the indicator species that comprise them, and environmental drivers that organize them at daily and inter-annual time scales. Finally, the differential responses in the diel, and inter-annual communities were used to provide evidence for a large-scale event that began to show an environmental signal in 2010 and subsided in 2011 and beyond. The event was manifested in the organization of the benthic fishes beginning weakly in 2010, peaking in 2011, and fully dissipating by 2012. The biotic effects of the event appeared to disproportionately affect the nighttime assemblage of fishes sampled on the WFS. Chapter Three explores the same WFS ecosystem, using the same fisheries-independent dataset, but also includes explicit modeling of the spatial variability captured by the sampling program undertaking the annual monitoring effort. The results also provided evidence of a disturbance that largely affected the nighttime fish community, and which was operating at spatial scales of variability that were larger than the extent of the shelf system itself. Like the previous study, the timing of this event is coincident with the 2010 Deepwater Horizon oil spill, the subsequent sub-marine dispersal of pollutants, and the cessation of spillage. Furthermore, the spatial models uncovered the influence of known spatial-abiotic gradients within the Gulf LME related to (1) depth, (2) temperature, and (3) salinity on the organization of daytime groundfish communities. Finally, the models developed also described which non-spatially structured abiotic variables were important to the observed beta-diversity. The ultimate results were the decomposition of the biotic response, within years and divided by diel classification, into the (1) pure-spatial, (2) pure-abiotic, (3) spatial-abiotic, and (4) unexplained fractions of variation. This study, along with that in Chapter Two, also highlighted the relative importance of the nighttime fish community to the assessment of the structure and function of the WFS, and the challenges associated with adequately sampling it, both in space and time. Because one focus of this dissertation was to develop low-decision frameworks and mathematically defensible alternatives to some common methods in fisheries ecology, Chapter Five employs a clustering technique to identify regime states that relies on hypothesis testing and the use of resemblance profiles as decision criteria. This clustering method avoids some of the arbitrary nature of common clustering solutions seen in ecology, however, it had never been rigorously subjected to numerical data simulation studies. Therefore, a formal investigation of the functional limits of the clustering method was undertaken prior to its use on real fisheries monitoring data, and is presented in Chapter Four. The results of this study are a set of recommendations for researchers seeking to utilize the new method, and the advice is applied in a case study in Chapter Five. Chapter Five presents the ecosystem-level fisheries indicator selection heuristic (EL-FISH) framework for examining long-term time-series data based on ecological monitoring for resources management. The focus of this study is the Gulf LME, encompassing the period of 1980-2011, and it specifically sought to determine to what extent the natural and anthropogenic induced environmental variability, including fishing extractions, affected the structure, function, and status of marine fisheries resources. The methods encompassed by EL-FISH, and the resulting ecosystem model that accounted for ~73% of the variability in biotic resources, allowed for (1) the identification and description of three fisheries resource regime state phase shifts in time, (2) the determination of the effects of fishing and environmental pressures on resources, and (3) providing context and evidence for trade-offs to be considered by managers and stakeholders when addressing fisheries management concerns. The EL-FISH method is fully transferrable and readily adapts to any set of continuous monitoring data.

Public health in Brazil presents many difficulties, and within this challenging context the University Hospital of Santa Maria (HUSM) RS, Brazil makes an effort to maintain its organization by keeping the focus on the efficiency of the healthcare services offered. Based on this fact, the present research aimed to investigate the number of appointments made at the ambulatories of HUSM during the years 2003 and 2005 using multivariate statistical techniques. Specific health areas, which offered ambulatory services, were selected based on the highest number of appointments for patients from the central region of Rio Grande do Sul State. Clustering and factorial analysis, carried out through the principal components, were multidimensional techniques used as a way to detect any relationship in the group of specific clinical areas considered in the study as well as the city where the patients came from. This way, the profile of the ambulatory services at HUSM, which are also offered by specialized clinics, is presented. The findings showed a significant number of patient sent to clinical surgery as well as to treatment with medication or supportive treatment. Among the 33 cities selected, 20 presented a more significant impact regarding appointments at the ambulatories. Therefore, cluster analysis and factorial analysis carried out through the principal components helped to show a relationship between the appointments of specific clinical areas at the ambulatories and the cities of the central region of the State of Rio Grande do Sul, providing support for the improvement of the quality of these services. / As dificuldades que envolvem a saúde pública no Brasil são muitas, e neste contexto de desafios diários está o HUSM, que luta para manter a organização sem perder de vista a eficiência na prestação de serviços. Com base nesta constatação a presente pesquisa trata da investigação sobre o número das consultas realizadas nos ambulatórios do Hospital Universitário de Santa Maria, no período de 2003 a 2005, através de técnicas estatísticas multivariadas. Para tanto, considerar-se-á os serviços ambulatoriais realizados por especialidades selecionadas com maior número de atendimentos para usuários da região central do Estado do Rio Grande do Sul. A Análise de Agrupamentos e Análise Fatorial através de componentes principais, foram técnicas multidimensionais utilizadas como meio de detectar relações no conjunto das especialidades clínicas consideradas na pesquisa, bem como o município de origem dos pacientes. Assim, revela-se o perfil dos atendimentos ambulatoriais do HUSM, os quais são oferecidos pelas clínicas especializadas. Indicando um número significativo de encaminhamentos para a clínica cirúrgica, como também para tratamento com medicação ou tratamento de apoio. Dentre os 33 municípios selecionados, 20 tiveram maior significância em relação às consultas ambulatoriais realizadas. Desse modo, a análise de agrupamentos e a análise fatorial através das componentes principais contribuíram indicando relações entre as consultas ambulatoriais das especialidades clínicas e os municípios da região central do Estado, fornecendo subsídios para a melhoria da qualidade destes serviços.

The quest for a complete understanding of mixtures is a challenge which has stimulated the development of several techniques. One of the most powerful NMR-based techniques is known as Diffusion-Ordered SpectroscopY (DOSY), in which it is possible to distinguish the NMR spectra of chemical species with different hydrodynamic radii, i.e. with different self-diffusion coefficients. It allows the study of intact mixtures, providing information on the interactions within the mixture and saving time and money compared to other techniques. Unfortunately, DOSY is not very effective when signals overlap and/or the diffusion coefficients are very similar. This drawback has led to the development of new methods to overcome this problem. The present investigation is focused on developing some of these. Most DOSY datasets show multiplet phase distortions caused by J-modulation. These distortions not only hinder the interpretation of spectra, but also increase the overlap between signals. The addition of a 45º purging pulse immediately before the onset of acquisition is proposed as a way to remove the unwanted distortions. Most DOSY experiments use 1H detection, because of the higher sensitivity which is generally achieved. However, acquiring spectra with other nuclei such as 13C can reduce overlap problems. Two new sequences have been developed to maximize the sensitivity of heteronuclear DOSY experiments. In order to increase resolving power, it is also possible to incorporate another variable into diffusion experiments as a further dimension. If this results in an approximately trilinear dataset (each dimension varying independently), it is possible to extract physically meaningful information for each component using multivariate statistical methods. This is explored for the cases where the new variable is either the relaxation behaviour or the concentration variation (which can be measured during a reaction or in a set of samples with different concentrations for each component). PARAllel FACtor (PARAFAC) analysis can obtain the spectra, diffusional decay and relaxation evolution or kinetics for each of the components. In a completely different approach, the separation power of liquid chromatography has been combined in a novel way with the NMR potential for elucidating structures. NMR has been used previously as a precise way to measure average flow velocities, even in porous media. Using this capability to detect the different average velocities of solutes that occur in chromatographic columns ought to provide a new way of analysing mixtures with the same potential as LC-NMR, but faster and more simple. In such a flow system, a chromatographic column is introduced into the NMR probe and a 2D dataset is acquired and Fourier transformed to obtain the velocity distribution for each of the detected NMR signals.

Craters formed by impact and volcanic processes are among the most fundamental planetary landforms. This study examines the morphology of diverse craterforms on Io, the Moon, Mars, and Earth using quantitative, outline-based shape analysis and multivariate statistical methods to evaluate the differences between different types of. Ultimately, this should help establish relationships between the form and origin of craterforms. Developed in the field of geometric morphometrics by paleontological and biological sciences communities, these methods were used for the analysis of the shapes of crater outlines. The shapes of terrestrial ash-flow calderas, terrestrial basaltic shield calderas, martian calderas, Ionian paterae, and lunar impact craters were quantified and compared. Specifically, we used circularity, ellipticity, elliptic Fourier analysis (EFA), Zahn and Roskies (Z-R) shape function, and diameter. Quantitative shape descriptors obtained from EFA yield coefficients from decomposition of the Fourier series that separates the vertical and horizontal components among the outline points for each shape. The shape descriptors extracted from Z-R analysis represent the angular deviation of the shapes from a circle. These quantities were subjected to multivariate statistical analysis including principal component analysis (PCA) and discriminant analysis, to examine maximum differences between each a priori established group. Univariate analyses of morphological quantities including diameter, circularity, and ellipticity, as well as multivariate analyses of elliptic Fourier coefficients and Z-R shape function angular quantities show that ash-flow calderas and paterae on Io, as well as basaltic shield calderas and martian calderas, are most similar in shape. Other classes of craters are also shown to be statistically distinct from one another. Multivariate statistical models provide successful classification of different types of craters. Three classification models were built with overall successful classification rates ranging from 90% to 75%, each conveying different shape information. The EFA model including coefficients from the 2nd to 10th harmonic was the most successful supervised model with the highest overall classification rate and most successful predictive group membership assignments for the population of examined craterforms. Multivariate statistical methods and classification models can be effective tools for analyzing landforms on planetary surfaces and geologic morphology. With larger data sets used to enhance supervision of the model, more successful classification by the supervised model could likely reveal clues to the formation and variables involved in the genesis of landforms.

At present, the East Asian Monsoon (EAM) influences water availability for nearly one third of the global population. The intensity and position of the EAM has varied considerably since its onset, but disagreement still exists related to the precise latitudinal and intensity shifts of the Westerly Jet and associated storm fronts, which mark the northern extent of the monsoon. Paleoclimate research can assist in improved assessment and prediction of EAM intensity, radiative forcing, and biogeochemical cycles in the Japan Sea and North Pacific, especially under the currently changing climate. My research primarily focuses on using major-, trace- and rare earth elements in sediments from International Ocean Drilling Program Expedition 346 in the Japan (Ulleung Basin) and East China Seas (Okinawa Trough) to track variability in the EAM on millennial time scales. Using geochemical and multivariate statistical techniques (Q-Mode Factor analysis and Constrained Least Squares multiple linear regressions), I differentiated compositionally similar terrigenous aluminosilicate materials (continental crust components, eolian dusts, volcanic ash) from these sediment archives. I successfully constructed a robust record of aluminosilicate provenance, which enables more precise determinations of EAM position and intensity than previously possible. Most of my research focused on the interpretation of aluminosilicate records over several different timescales from three sites from Expedition 346. In tandem with this research, I also refined values of the well-known, and widely used, Standard Reference Material (SRM) Hawaiian Volcano Observatory Basalt (BHVO-2). In the Okinawa trough (Sites U1428/U1429), I identified and tracked the increase in flux of five continental crust materials, loesses, and volcanic ashes during glacial cycles, continental shelf exposure, and the migration of paleo-rivers in the last 400 kyr. Additionally, I constructed a 12 Myr record, which identified and quantified the dust fluxes to Ulleung Basin (Site U1430), and emphasized the importance of the Taklimakan and Gobi Deserts as main sources of dust to the Japan Sea and Pacific through the Cenozoic. Collectively, these aluminosilicate flux reconstructions are first to identify multiple specific Asian source regions through the Cenozoic, and highlight the complexity of accurately reconstructing monsoons and other aspects of paleoclimate from sediment in dynamic environments.

In this work, an automated statistical approach for the condition monitoring of a fluid power system based on a process sensor network is presented. In a multistep process, raw sensor data are processed by feature extraction, selection and dimensional reduction and finally mapped to discriminant functions which allow the detection and quantification of fault conditions. Experimentally obtained training data are used to evaluate the impact of temperature and different aeration levels of the hydraulic fluid on the detection of pump leakage and a degraded directional valve switching behavior. Furthermore, a robust detection of the loading state of the installed filter element and an estimation of the particle contamination level is proposed based on the same analysis concept.

In this thesis we derive the distribution for the first antieigenvalue for a random matrix with distribution W ∼ Wp(n, Ip) for p = 2 and p = 3. For p = 2 we present a proof that the first antieigenvalue has distribution β((n−1)/2, 1). For p = 3 we prove that the probability density function can be expressed using a sum of hypergeometric functions. Besides the main objective, the thesis seeks to introduce the theory of multivariate statistics and antieigenvalues.