As researchers increasingly use web-based surveys, the ease of dropping out in the online setting is a growing issue in ensuring data quality. One theory is that dropout or attrition occurs in phases that can be generalized to phases of high dropout and phases of stable use. In order to detect these phases, several methods are explored. First, existing methods and user-specified thresholds are applied to survey data where significant changes in the dropout rate between two questions is interpreted as the start or end of a high dropout phase. Next, survey dropout is considered as a time-to-event outcome and tests within change-point hazard models are introduced. Performance of these change-point hazard models is compared. Finally, all methods are applied to survey data on patient cancer screening preferences, testing the null hypothesis of no phases of attrition (no change-points) against the alternative hypothesis that distinct attrition phases exist (at least one change-point).

This dissertation is an evaluation of the generalization characteristics of machine learning classifiers as applied to the detection of coral reefs using remote sensing images. Three scientific studies have been conducted as part of this research: 1) Evaluation of Spatial Generalization Characteristics of a Robust Classifier as Applied to Coral Reef Habitats in Remote Islands of the Pacific Ocean 2) Coral Reef Change Detection in Remote Pacific Islands using Support Vector Machine Classifiers 3) A Generalized Machine Learning Classifier for Spatiotemporal Analysis of Coral Reefs in the Red Sea. The aim of this dissertation is to propose and evaluate a methodology for developing a robust machine learning classifier that can effectively be deployed to accurately detect coral reefs at scale. The hypothesis is that Landsat data can be used to train a classifier to detect coral reefs in remote sensing imagery and that this classifier can be trained to generalize across multiple sites. Another objective is to identify how well different classifiers perform under the generalized conditions and how unique the spectral signature of coral is as environmental conditions vary across observation sites. A methodology for validating the generalization performance of a classifier to unseen locations is proposed and implemented (Controlled Parameter Cross-Validation,). Analysis is performed using satellite imagery from nine different locations with known coral reefs (six Pacific Ocean sites and three Red Sea sites). Ground truth observations for four of the Pacific Ocean sites and two of the Red Sea sites were used to validate the proposed methodology. Within the Pacific Ocean sites, the consolidated classifier (trained on data from all sites) yielded an accuracy of 75.5% (0.778 AUC). Within the Red Sea sites, the consolidated classifier yielded an accuracy of 71.0% (0.7754 AUC). Finally, long-term change detection analysis is conducted for each of the sites evaluated. In total, over 16,700 km2 was analyzed for benthic cover type and cover change detection analysis. Within the Pacific Ocean sites, decreases in coral cover ranged from 25.3% reduction (Kingman Reef) to 42.7% reduction (Kiritimati Island). Within the Red Sea sites, decrease in coral cover ranged from 3.4% (Umluj) to 13.6% (Al Wajh).

The focus of this thesis was to investigate which baseball metrics are most conducive to run creation and prevention. Stepwise regression and Liu estimation were used to formulate two models for the dependent variables and also used for cross validation. Finally, the predicted values were fed into the Pythagorean Expectation formula to predict a team’s most important goal: winning. Each model fit strongly and collinearity amongst offensive predictors was considered using variance inflation factors. Hits, walks, and home runs allowed, infield putouts, errors, defense-independent earned run average ratio, defensive efficiency ratio, saves, runners left on base, shutouts, and walks per nine innings were significant defensive predictors. Doubles, home runs, walks, batting average, and runners left on base were significant offensive regressors. Both models produced error rates below 3% for run prediction and together they did an excellent job of estimating a team’s per-season win ratio.

In this dissertation, we have developed and combined several statistical techniques in Bayesian factor analysis (BAYFA) and mixture of factor analyzers (MFA) to overcome the shortcoming of these existing methods. Information Criteria are brought into the context of the BAYFA model as a decision rule for choosing the number of factors m along with the Press and Shigemasu method, Gibbs Sampling and Iterated Conditional Modes deterministic optimization. Because of sensitivity of BAYFA on the prior information of the factor pattern structure, the prior factor pattern structure is learned directly from the given sample observations data adaptively using Sparse Root algorithm. Clustering and dimensionality reduction have long been considered two of the fundamental problems in unsupervised learning or statistical pattern recognition. In this dissertation, we shall introduce a novel statistical learning technique by focusing our attention on MFA from the perspective of a method for model-based density estimation to cluster the high-dimensional data and at the same time carry out factor analysis to reduce the curse of dimensionality simultaneously in an expert data mining system. The typical EM algorithm can get trapped in one of the many local maxima therefore, it is slow to converge and can never converge to global optima, and highly dependent upon initial values. We extend the EM algorithm proposed by cite{Gahramani1997} for the MFA using intelligent initialization techniques, K-means and regularized Mahalabonis distance and introduce the new Genetic Expectation Algorithm (GEM) into MFA in order to overcome the shortcomings of typical EM algorithm. Another shortcoming of EM algorithm for MFA is assuming the variance of the error vector and the number of factors is the same for each mixture. We propose Two Stage GEM algorithm for MFA to relax this constraint and obtain different numbers of factors for each population. In this dissertation, our approach will integrate statistical modeling procedures based on the information criteria as a fitness function to determine the number of mixture clusters and at the same time to choose the number factors that can be extracted from the data.

The Multi-Agency Radiological Site Survey Investigation Manual (MARSSIM) is a regulatory guidance document regarding compliance evaluation of radiologically contaminated soils and buildings (USNRC, 2000). Compliance is determined by comparing radiological measurements to established limits using a combination of hypothesis testing and scanning measurements. Scanning allows investigators to identify localized pockets of contamination missed during sampling and allows investigators to assess radiological exposure at different spatial scales. Scale is important in radiological dose assessment as regulatory limits can vary with the size of the contaminated area and sites are often evaluated at more than one scale (USNRC, 2000). Unfortunately, scanning is not possible in the subsurface and direct application of MARSSIM breaks down. This dissertation develops a subsurface decision framework called the Geospatial Extension to MARSSIM (GEM) to provide multi-scale subsurface decision support in the absence of scanning technologies. Based on geostatistical simulations of radiological activity, the GEM recasts the decision rule as a multi-scale, geospatial decision rule called the regulatory limit rule (RLR). The RLR requires simultaneous compliance with all scales and depths of interest at every location throughout the site. The RLR is accompanied by a compliance test called the stochastic conceptual site model (SCSM). For those sites that fail compliance, a remedial design strategy is developed called the Multi-scale Remedial Design Model (MrDM) that spatially indicates volumes requiring remedial action. The MrDM is accompanied by a sample design strategy known as the Multi-scale Remedial Sample Design Model (MrsDM) that refines this remedial action volume through careful placement of new sample locations. Finally, a new sample design called “check and cover” is presented that can support early sampling efforts by directly using prior knowledge about where contamination may exist. This dissertation demonstrates how these tools are used within an environmental investigation and situates the GEM within existing regulatory methods with an emphasis on the Environmental Protection Agency’s Triad method which recognizes and encourages the use of advanced decision methods. The GEM is implemented within the Spatial Analysis and Decision Assistance (SADA) software and applied to a hypothetical radiologically contaminated site.

Copy-number variation (CNV) has been implicated in many complex diseases. It is of great interest to detect and locate such regions through genetic association testings. However, the association testings are complicated by the fact that CNVs usually span multiple markers and thus such markers are correlated to each other. To overcome the difficulty, it is desirable to pool information across the markers. In this thesis, we propose a kernel-based method for aggregation of marker-level tests, in which first we obtain a bunch of p-values through association tests for every marker and then the association test involving CNV is based on the statistic of p-values combinations. In addition, we explore several aspects of its implementation. Since p-values among markers are correlated, it is complicated to obtain the null distribution of test statistics for kernel-base aggregation of marker-level tests. To solve the problem, we develop two proper methods that are both demonstrated to preserve the family-wise error rate of the test procedure. They are permutation based and correlation base approaches. Many implementation aspects of kernel-based method are compared through the empirical power studies in a number of simulations constructed from real data involving a pharmacogenomic study of gemcitabine. In addition, more performance comparisons are shown between permutation-based and correlation-based approach. We also apply those two approaches to the real data. The main contribution of the dissertation is the development of marker-level association testing, a comparable and powerful approach to detect phenotype-associated CNVs. Furthermore, the approach is extended to high dimension setting with high efficiency.

This paper intend to find the proper hypothesis and test statistic for testing existence of bilaterally contamination when there exists nuisance parameter. The test statistic is based on method of moments estimators. Union-Intersection test is used for testing if the distribution of population can be implemented by a bilaterally contaminated normal model with unknown variance. This paper also developed a hierarchical normal mixture model (HNM) and applied it to birth weight data. EM algorithm is employed for parameter estimation and a singular Bayesian information criterion (sBIC) is applied to choose the number components. We also proposed a singular flexible information criterion which in addition involves a data-driven penalty.

Submitted by Claudia Rocha (claudia.rocha@udesc.br) on 2017-12-08T11:26:22Z No. of bitstreams: 1 PGEF16MA058.pdf: 2613075 bytes, checksum: 6818848cc42688c9ae9c287f102d82f5 (MD5) / Made available in DSpace on 2017-12-08T11:26:22Z (GMT). No. of bitstreams: 1 PGEF16MA058.pdf: 2613075 bytes, checksum: 6818848cc42688c9ae9c287f102d82f5 (MD5) Previous issue date: 2016-02-22 / FUMDES / This work aimed to evaluate nineteen tapering models divided into five groups as the framework for the estimation of the profile, relative height and volume on a planted forest of Araucaria angustifolia (Bertol.) Kuntze. The settlements are located in the West of the State of Santa Catarina, deployed in the 1970 and 1980. The trees were divided into classes of diameter and height, by means of this stratification divided the data into two groups being 70% used for adjustment and 30% used for validation of the equations. In the adjustment of the models the best statistics were for the model described by Bi (2000), showing how R²aj. 0,9852, Standard Error of Estimate Relative (%) value of Syx 6,95% and Akaike information criterion (AIC) 964.5 value. Based on the graphical analysis of waste, statistics of the deviation (D), Sum of the square of the relative error (SQER) and residue in percentage (RP), it was found that for the estimation of the diameter along the stem form equations variable obtained the best settings, and the equation of Lee et al. (2003), obtained from value D 6,8361, SRRP 0,0796 end RP 0,3338. To estimate the heights in different diameters equation that best estimated Heights was the equation developed by Kozak (1988), as the value of D 0,1471, SRRQ 0,1457 and RP 4,9396. To estimate the volumes of mergers the equation that stood out was described by Lee et al. (2003), so variable, showing value of D 0,0004, SRRQ 0,2318 and RP 0,0028. Therefore, the estimates of the diameters, heights and the group volume of the equations that best stood out were those of variable shape and equation which best presented the results of estimates for the variables was the equation proposed by Lee et al. (2003). The Chi-square test (χ ²) all the equations were able to perform estimates. To do so, were generated a graph and a table of assortment which can figure out the percentage of each product generated and also the absolute volume of wood in each diameter class / Este trabalho teve como objetivo avaliar dezenove modelos de afilamento divididos em cinco grupos quanto a estrutura, para à estimativa do perfil, altura relativa e volume em uma floresta plantada de Araucaria angustifolia (Bertol.) Kuntze. Os povoamentos estão localizados na região Oeste do estado de Santa Catarina, implantados nas décadas de 1970 e 1980. As árvores foram divididas em classes de diâmetro e de altura, por meio desta estratificação dividiu-se os dados em dois grupos sendo 70% utilizado para o ajuste e 30% utilizados para a validação das equações. No ajuste dos modelos as melhores estatísticas foram para o modelo descrito por Bi (2000), apresentando como R²aj. 0,9852, Erro Padrão da Estimativa Relativo (Syx%) o valor de 6,95% e Critério de Informação Akaike (AIC) o valor de 964,5. Com base na análise gráfica dos resíduos, na estatística do Desvio (D), Soma do Quadrado do Erro Relativo (SQER) e Resíduo em porcentagem (RP), verificou-se que para a estimativa dos diâmetros ao longo do fuste as equações de forma variável obtiveram os melhores ajustes, sendo que a equação de Lee et al. (2003), obteve valor de D de -0,0796, SQER 6,8361 e RP de -0,3338. Para estimativa das alturas em diferentes diâmetros a equação que melhor estimou as alturas foi a equação desenvolvida por Kozak (1988), tendo como valor de D -0,1471, SQER 0,1457 e RP -4,9396. Para estimativa dos volumes dos troncos a equação que se destacou foi a descrita por Lee et al. (2003), de forma variável, apresentando valor de D 0,0004, SQER 0,0028 e RP 0,2318. Portanto, para as estimativas dos diâmetros, alturas e volume o grupo das equações que melhor se destacaram foram as de forma variável e a equação que melhor apresentou os resultados de estimativas para as variáveis foi a equação proposta por Lee et al. (2003). No teste de qui-quadrado (χ²) todas as equações mostraram-se aptas a realizar as estimativas. Para tanto, foram gerados um gráfico e uma tabela de sortimento os quais pode-se perceber a porcentagem de cada produto gerado e também o volume absoluto de madeira em cada classe de diâmetro

Made available in DSpace on 2015-05-14T12:47:16Z (GMT). No. of bitstreams: 1 arquivototal.pdf: 12803738 bytes, checksum: 10bba02a1a8ff46e06bbe49fd4d6d729 (MD5) Previous issue date: 2013-03-15 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / Besides being important interpersonal communication mechanism, the voice also takes on the role of the main working tool for specific professions, such as teachers. And These occupations require high vocal demand that associated with certain kinds of factors may contribute to the increased prevalence vocal complaints at this population, limiting occupational activity and generating high financial and social costs. This study aimed To analyze the prevalence of vocal and biopsychosocial aspects associated with this disorder in teachers of municipal schools in the city of João Pessoa. To make it possible, there was a sectional and observational research, exploratory and descriptive analytic, between the months of September 2011 to March 2013. The final sample consisted of 270 professionally active teachers in elementary schools in the city of João Pessoa, coming from 21 schools. The subjects had s vocal sample collected by free software Praat (version 5.0.32) and answered four questionnaires (CPV-P, VAPP, STAI-trait, SRQ) aimed to assessing occupational factors, physical, social, emotional and psychological in the participants. The study was approved by the Department of Education and Culture of the city of João Pessoa and by the Ethics in Research from the CCS / UFPB. The software used for statistical analysis was the software livre R. To develop the decision model for dysphonia, it was used the Logistic Regression Model.To select the best model we used the AIC standard and to verify the validity and quality of the adjustment model we used the Hosmer-Lemeshow and Le-van Houwelingen Cessie and area under the ROC curve. The conclusions of the model were presented in odds ratio. The results showed that the prevalence of dysphonia in these professionals is 86.3% and that more than half of the sample (58.0%) presented change vocal mild to moderate. There is no consistence between the assessment by expert and the selfassessment made by the speaker himself, noting that, generally, the teacher tends to minimize his/her problem. High anxiety is present in 41.8% of the teachers and the presence of psychological distress was found in 30.0% of respondents. The logistic regression models for s risk factors and to the group variables physical, social, emotional and psychic related more dysphonia selected variables (respectively): Sex / weekly schedule / location for rest / loud noise / use voice in corner unprofessional practices and failures in voice / voice thin or acute / lack of appetite / difficulty in making decisions / not consider themselves more important than they think. The models could suggest a method of screening faster and more effective in large groups of teachers in order to make it easier the selection of people with higher risk of developing the vocal problem. However, this issue needs to be discussed more intensively in sectors responsible for worker health, including in our city where the actions directed to the health of professional teachers are scarce, limited and isolated. / Além de ser mecanismo de comunicação interpessoal importante, a voz assume também o papel de ferramenta principal de trabalho para determinadas categorias profissionais, como os professores. Essas ocupações exigem alta demanda vocal que, associada a determinados fatores, podem contribuir para o aumento da prevalência de queixas vocais nestes grupos de profissionais, limitando a atividade ocupacional e gerando altos custos financeiros e sociais. Este trabalho teve como objetivo analisar a prevalência de alteração vocal e os aspectos biopsicossociais associados a esse transtorno em professores da rede municipal de ensino da cidade de João Pessoa. Para isto, realizou-se uma pesquisa observacional e seccional, de natureza exploratória descritiva e analítica, que compreenderam os meses de setembro de 2011 a março de 2013. A amostra final foi composta por 270 professores profissionalmente ativos no ensino fundamental da cidade de João Pessoa, advindos de 21 unidades escolares. Os sujeitos tiveram amostras vocais coletadas pelo software livre Praat (versão 5.0.32) e responderam a quatro questionários (CPV-P, PPAV, IDATE-traço, SRQ) voltados à avaliação de fatores ocupacionais, físicos, sociais, emocionais e psíquicos dos participantes. A pesquisa foi aprovada pela Secretaria de Educação e Cultura do município de João Pessoa e pelo Comitê de Ética em Pesquisa do CCS/UFPB. O software utilizado para as análises estatísticas foi o software livre R. Para desenvolvimento do modelo de decisão para a disfonia, utilizou-se o Modelo de Regressão Logística. Para seleção do melhor modelo foi utilizado o critério AIC e para a verificação da validade e qualidade do ajuste do modelo foram utilizados o teste de Hosmer e Lemeshow e Le Cessie-van Houwelingen e a área sob a curva ROC. As conclusões do modelo foram apresentadas em forma de Probabilidades e Razão de Chances (Odds Ratio). Os resultados mostraram que a prevalência da disfonia nestes profissionais é de 86,3% e que mais da metade da amostra (58,0%) apresenta alteração vocal de grau leve a moderado. Não há compatibilidade entre a avaliação realizada pelo especialista e a autoavaliação feita pelo próprio falante, observando-se que, geralmente, o professor tende a minimizar seu problema. A alta ansiedade está presente em 41,8% dos professores e a presença de distúrbios psíquicos menores foi encontrada em 30,0% dos entrevistados. Os modelos de regressão logísticos para os fatores de risco e para o grupo de variáveis físicas, sociais, emocionais e psíquicas mais relacionadas com a disfonia selecionaram as variáveis (respectivamente): sexo/carga horária semanal/local para descanso/ruído forte/uso da voz em práticas de canto não profissionais e falhas na voz/voz fina ou aguda/falta de apetite/dificuldades em tomar decisões/não se autoconsiderar mais importante do que pensam. Os modelos puderam sugerir um método de triagem mais rápido e eficaz em grandes grupos de professores, visando facilitar a seleção de indivíduos com maior risco de apresentarem ou desenvolverem o problema vocal. Contudo, esta temática precisa ser mais intensivamente discutida nos setores responsáveis pela saúde do trabalhador, inclusive em nosso município, em que as ações direcionadas à saúde do profissional docente ainda são escassas, limitadas e isoladas.

L'adaptation locale est un phénomène micro-évolutif qui peut survenir lorsque des populations d'une même espèce sont exposées à des conditions environnementales différentes.Si cet environnement exerce une pression sous forme de sélection naturelle, qu'il existe un potentiel adaptatif au sein des populations et que le flux de gènes est suffisamment modéré, les populations vont alors tendre vers un optimum adaptatif local.Dans cette thèse, je m'intéresse aux moyens méthodologiques de l'étude de l'adaptation locale d'une part, et à l'étude de ce phénomène le long d'un gradient d'altitude chez la plante alpine Arabis alpina d'autre part.Dans la première partie méthodologique, je montre que les méthodes de scan génomique pour détecter les marqueurs génétiques sous sélection peuvent souffrir de forts taux de faux positifs lorsqu'exposées à des jeux de données complexes, mais réalistes.Je présente ensuite une méthode statistique de détection de marqueurs génétiques sous sélection qui, contrairement aux méthodes existantes, utilisent à la fois la notion de différentiation génétique (ou Fst) et une information environnementale.Cette méthode a été développée de manière à limiter son taux de faux positifs de manière générale.J'offre enfin une perspective concernant les liens entre une expérience ancienne en biologie évolutive (l'expérience de jardin commun) et les nouveaux développements moléculaires et statistiques modernes.Dans la seconde partie empirique, je présente une analyse de la démographie d'A. alpina dans six populations naturelles. Outre qu'elle révèle des caractéristiques biologiques intéressantes sur cette espèce (faible espérance de vie, reproduction et survie très différentielle...), cette analyse montre que la croissance diminue et la survie augmente chez cette espèce avec la baisse de la température moyenne (donc avec l'altitude).Puisque ces analyses ne permettent pas d'exclure des hypothèses de dérive et de plasticité phénotype, je présente une analyse en jardin commun sur A. alpina qui permet de lisser les problèmes de plasticité phénotypique et qui, combinée à des analyses moléculaires, permettent d'exclure l'hypothèse de dérive.Les résultats montrent qu'il existe un syndrome phénotypique adaptatif lié à la température moyenne qui tend à des plantes plus petites, plus compactes, qui croissent et se reproduisent moins, dans les milieux froids.À l'aide des données moléculaires et de méthodes de scan génomique, je présente une liste de 40 locus qui peuvent être impliqués dans ce processus.Pour finir, je discute l'ensemble de ces résultats empiriques dans un contexte plus général d'écologie alpine. Je résume ensuite les principaux obstacles méthodologiques à l'étude de l'adaptation locale et je fourni quelques perspectives méthodologiques. / Local adaptation is a micro-evolutionary phenomenon, which arises when populations of the same species are exposed to contrasted environmental conditions.If this environment exert some natural selection pressure, if an adaptive potential exists among the populations and if the gene flow is sufficiently mild, populations are expected to tend toward a local adaptive optimum.In this thesis, I study the methodological means of the study of local adaptation on the one hand, and I investigate this phenomenon along an elevation gradient in the alpine plant Arabis alpina on the other hand.In the first, methodological part, I show that the genome scan methods to detect selection using genetic markers might suffer strong false positive rates when confronted to complex but realistic datasets.I then introduce a statistical method to detect markers under selection, which, contrary to existing methods, make use of both the concept of genetic differentiation (or Fst) and environmental information.This method has been developed in order to reduce its global false positive rate.Finally, I present some perspectives regarding the relationships between the relatively old ``common garden'' experiment and the new developments in molecular biology and statistics.In the second, empirical part, I introduce an analysis of the demographic characteristics of A. alpina in six natural populations. Besides providing interesting biological information on this species (low life expectancy, strongly contrasted reproduction and survival...), these analyses show that growth increase and survival decrease with the decrease of average temperature (hence with altitude).Since these analyses do not allow us to rule out hypotheses such as drift and phenotypic plasticity, I show the results of a common garden experiment which enable us to smooth phenotypic plasticity and, when combined with molecular data, enable us to rule out the hypothesis of drift.The results show the existence of an adaptive phenotypic syndrome, in which plants are smaller, are more compact, grow slower and reproduce less in cold temperature environments.Using the molecular data, I draw a list of 40 locus which might be involved in this adaptive process.In the end, I discuss these empirical findings as a whole to place them in a more general context of alpine ecology. I sum up the main methodological challenges when studying local adaptation and offer some methodological perspectives.