La question du rachat préoccupe les assureurs depuis longtemps notamment dans le contexte des contrats d'épargne en Assurance-Vie, pour lesquels des sommes colossales sont en jeu. L'émergence de la directive européenne Solvabilité II, qui préconise le développement de modèles internes (dont un module entier est dédié à la gestion des risques de comportement de rachat), vient renforcer la nécessité d'approfondir la connaissance et la compréhension de ce risque. C'est à ce titre que nous abordons dans cette thèse les problématiques de segmentation et de modélisation des rachats, avec pour objectif de mieux connaître et prendre en compte l'ensemble des facteurs-clefs qui jouent sur les décisions des assurés. L'hétérogénéité des comportements et leur corrélation ainsi que l'environnement auquel sont soumis les assurés sont autant de difficultés à traiter de manière spécifique afin d'effectuer des prévisions. Nous développons ainsi une méthodologie qui aboutit à des résultats très encourageants ; et qui a l'avantage d'être réplicable en l'adaptant aux spécificités de différentes lignes de produits. A travers cette modélisation, la sélection de modèle apparaît comme un point central. Nous le traitons en établissant les propriétés de convergence forte d'un nouvel estimateur, ainsi que la consistance d'un nouveau critère de sélection dans le cadre de mélanges de modèles linéaires généralisés / Insurers have been concerned about surrenders for a long time especially in Saving business, where huge sums are at stake. The emergence of the European directive Solvency II, which promotes the development of internal risk models (among which a complete unit is dedicated to surrender risk management), strengthens the necessity to deeply study and understand this risk. In this thesis we investigate the topics of segmenting and modeling surrenders in order to better know and take into account the main risk factors impacting policyholders’ decisions. We find that several complex aspects must be specifically dealt with to predict surrenders, in particular the heterogeneity of behaviours and their correlations as well as the context faced by the insured. Combining them, we develop a methodology that seems to provide good results on given business lines, and that moreover can be adapted for other products with little effort. However the model selection step suffers from a lack of parsimoniousness: we suggest to use another criteria based on a new estimator, and prove its consistant properties in the framework of mixtures of generalized linear models

Master's thesis focuses on processing fMRI data, which are mapping blood oxygenation level dependence in a state of brain activity. Usable and necessarily preprocessing tech- niques of the data, together with two main analysis approaches are introduced. The area of univariate methods, especially general linear model and multivariate principal or independent component analysis is explained. Practical application of the methods involved on the real fMRI data set is implemented. Relevant results as well as theirs mutual possible comparison is presented.

Submitted by Jean Medeiros (jeanletras@uepb.edu.br) on 2016-08-25T12:54:34Z No. of bitstreams: 1 PDF - Thainá Alves Lycarião.pdf: 852519 bytes, checksum: 636ecf5fa1e6e551257df1fb958919b3 (MD5) / Made available in DSpace on 2016-08-25T12:54:34Z (GMT). No. of bitstreams: 1 PDF - Thainá Alves Lycarião.pdf: 852519 bytes, checksum: 636ecf5fa1e6e551257df1fb958919b3 (MD5) Previous issue date: 2015-02-24 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / Understanding the dynamics of ecosystems in semiarid regions with dry and wet extremes may help elucidate issues related to the existing pattern of interaction in these environments, and these relationships can affect the ecological interactions of organisms present in the water body. Thus, this study aimed to determine the abiotic variables that act in development of Najas arguta (Kunth) and Egeria densa (Planch), highlighting how these variables influence the relationship of the two submerged macrophytes in a reservoir in semiarid region. N arguta biomass data and E. densa were obtained in six stands, located three out of the reservoir shoreline. In each stands is selected parcels 100x50 meters, which have been measured abiotic data and collected plants through cross hooks (cor er ) 0.5m in diameter (n = 4) at depths 1m, 2m and 3m. Analysis of Variance (ANOVA) factor a + b was used to test the seasonal and spatial variation of stands and environmental variables, linear regression analysis was used to understand the dependence of N. arguta and E. densa . Generalized Linear Models (GLM) were used to select environmental variables that explain the occurrence of N. arguta , E. densa and its relationship (Najas:Egeria ). Abiotic data showed low coefficients of temporal variation (<30%) due to continued drought conditions. N. Arguta presented a variation of biomass between the months of August / 13 and April / 14, with totals of 25.67 g.m-² and 1.06 g.m-², respectively. While E. densa maintained high biomass in the entire study period. The submerged plants had different niches, showing that E. densa competition relationship with N. arguta . The predictive model for biomass of N. arguta showed that physical and chemical variables had more influence in their biomass, whereas physical, chemical and climatic variables were related more to the biomass of E. densa . The relationship Najas Egeria was influenced by conductivity, total dissolved solids and nitrite. For presenting adaptations to the conditions of the reservoir, E. densa kept his high biomass, occupying more favorable niches and suppressing the development of N. arguta . / Entender a dinâmica dos ecossistemas em regiões do semiárido que apresentam extremos de seca e cheia pode ajudar a elucidar questões relacionadas ao padrão de interação existentes nesses ambientes, bem como essas relações podem afetar nas interações ecológicas dos organismos presentes no corpo aquático. Desta forma, o presente trabalho teve como objetivo determinar as variáveis abióticas que atuam no desenvolvimento de Najas arguta (Kunth) e Egeria densa (Planch), ressaltando como estas variáveis influenciam na relação das duas macrófitas submersas em um reservatório no semiárido. Dados de biomassa de N. arguta e E. densa foram obtidos em seis bancos, localizados três em cada margem do reservatório. Em cada selecionou-se parcelas 100x50 metros, onde foram medidos dados abióticos e coletados plantas através de ganchos de 0,5m de diâmetro (n=4), nas profundidades 1m, 2m e 3m. banco Análise de Variância (ANOVA) fatorial a+b foi utilizada para testar a variação sazonal e espacial dos bancos e das variáveis ambientais, análise de regressão linear foi utilizada para entender a relação de dependência de N. arguta e E. densa . Modelos Lineares Generalizados (GLM) foram utilizados para selecionar as variáveis ambientais que explicam a ocorrência de N. arguta , E. densa e sua relação (Naja s:Egeria ). Os dados abióticos apresentaram baixo coeficiente de variação temporal (<30%) devido à manutenção da condição de estiagem. N. arguta apresentou variação de biomassa entre - - os meses de agosto/13 e abril/14, com valores totais de 25.67 g.m ² e 1.06 g.m ², respectivamente. Enquanto que E. densa manteve biomassa alta em todo período estudado. As macrófitas submersas apresentaram nichos diferenciados, mostrando haver uma competição de E. densa em relação N. arguta . O modelo preditivo para a biomassa de N. arguta mostrou que variáveis físicas e químicas tiveram mais influencia na sua biomassa, enquanto que variáveis físicas, químicas e climatológicas se relacionaram mais com a biomassa de E. densa . A relação de Najas:Egeria foi influenciada pela condutividade, sólidos totais dissolvidos e nitrito. Por apresentar adaptações condições do reservatório, E. densa manteve sua biomassa alta, ocupando nichos mais às favoráveis e suprimindo o desenvolvimento de N. arguta.

My Masters research aims to deepen our understanding of the behaviour of robust methods in logistic regression. Logistic regression is a special case of Generalized Linear Modelling (GLM), which is a powerful and popular technique for modelling a large variety of data. Robust methods are useful in reducing the effect of outlying values in the response variable on parameter estimates. A literature survey shows that we are still at the beginning of being able to detect extreme observations in logistic regression analyses, to apply robust methods in logistic regression and to present informatively the results of logistic regression analyses. In Chapter 1 I have made a basic introduction to logistic regression, with an example, and to robust methods in general. In Chapters 2 through 4 of the thesis I have described traditional methods and some relatively new methods for presenting results of logistic regression using powerful visualization techniques as well as the concepts of outliers in binomial data. I have used different published data sets for illustration, such as the Prostate Cancer data set, the Damaged Carrots data set and the Recumbent Cow data set. In Chapter 4 I summarize and report on the modem concepts of graphical methods, such as central dimension reduction, and the use of graphics as pioneered by Cook and Weisberg (1999). In Section 4.6 I have then extended the work of Cook and Weisberg to robust logistic regression. In Chapter 5 I have described simulation studies to investigate the effects of outlying observations on logistic regression (robust and non-robust). In Section 5.2 I have come to the conclusion that, in the case of classical or robust multiple logistic regression with no outliers, robust methods do not necessarily provide more reasonable estimates of the parameters for the data that contain no st~ong outliers. In Section 5.4 I have looked into the cases where outliers are present and have come to the conclusion that either the breakdown method or a sensitivity analysis provides reasonable parameter estimates in that situation. Finally, I have identified areas for further study.

I investigated the role of extra-pair paternity on use of public information and the interaction between public information and personal information for patch fidelity decisions. It is unknown if songbirds use public information about the number of conspecific fledglings for patch fidelity decisions when extra-pair paternity is uncommon. I tested if probability of patch fidelity was associated with (1) number of fledglings in adjacent territories (public information), and (2) number of fledglings raised with a social mate (personal information). I used logistic regression to predict probability of patch fidelity of males and females based on the 2 uncorrelated predictor variables (Spearman’s rank correlation, S = 21895.28, n = 50, P = 0.723, r = –0.051). I monitored patch fidelity of 107 territories, counted the number of fledglings in each territory, and assessed parentage of 102 young from 36 nests for white-eyed vireos (Vireo griseus) from 2008–2010 in a 100 ha patch of woodland in central Texas, USA. I excluded the social male as the father of 3 of the 102 young and did not exclude any of the social females as the mother with parentage analysis using 6 microsatellite loci. The number of fledglings in adjacent territories was not a good predictor of probability of patch fidelity for males (beta 1 = 0.166, df = 35, P = 0.247, Nagelkerke’s R2 = 0.054) or females (beta 1 = 0.121, df = 17, P = 0.670, Nagelkerke’s R2 = 0.016). The number of fledglings raised with a social mate was also not a good predictor of probability of patch fidelity for males (beta 1 = –0.296, df = 43, P = 0.360, Nagelkerke’s R2 = 0.029), whereas it was a good predictor for females (beta 1 = 1.281, df = 21, P = 0.048, Nagelkerke’s R2 = 0.409). The dominant ecological concepts for explaining site fidelity in songbirds, win-stay lose-switch (based on personal reproductive success with a social mate) and public information, did not predict probability of patch fidelity well for male white-eyed vireos. The win-stay lose-switch model, but not public information was a good predictor of probability of patch fidelity for females. My results suggest that use of public information may depend on frequency of extra-pair paternity. Males may primarily use other information for patch fidelity decisions beyond reproductive success of conspecifics for patch fidelity decisions in some circumstances. My results support the need to ensure high levels of nesting success for females to return and maintain populations in areas managed for breeding songbirds for conservation efforts to be successful.

This research focuses on developing comprehensive frameworks for developing robust parameter design methodology for dynamic systems with automatic control and for synthesis of nanostructures. In many automatically controlled dynamic processes, the optimal feedback control law depends on the parameter design solution and vice versa and therefore an integrated approach is necessary. A parameter design methodology in the presence of feedback control is developed for processes of long duration under the assumption that experimental noise factors are uncorrelated over time. Systems that follow a pure-gain dynamic model are considered and the best proportional-integral and minimum mean squared error control strategies are developed by using robust parameter design. The proposed method is illustrated using a simulated example and a case study in a urea packing plant. This idea is also extended to cases with on-line noise factors. The possibility of integrating feedforward control with a minimum mean squared error feedback control scheme is explored. To meet the needs of large scale synthesis of nanostructures, it is critical to systematically find experimental conditions under which the desired nanostructures are synthesized reproducibly, at large quantity and with controlled morphology. The first part of the research in this area focuses on modeling and optimization of existing experimental data. Through a rigorous statistical analysis of experimental data, models linking the probabilities of obtaining specific morphologies to the process variables are developed. A new iterative algorithm for fitting a Multinomial GLM is proposed and used. The optimum process conditions, which maximize the above probabilities and make the synthesis process less sensitive to variations of process variables around set values, are derived from the fitted models using Monte-Carlo simulations. The second part of the research deals with development of an experimental design methodology, tailor-made to address the unique phenomena associated with nanostructure synthesis. A sequential space filling design called Sequential Minimum Energy Design (SMED) for exploring best process conditions for synthesis of nanowires. The SMED is a novel approach to generate sequential designs that are model independent, can quickly "carve out" regions with no observable nanostructure morphology, and allow for the exploration of complex response surfaces.

Maritime is a very competitive industry in global market. In addition to political, economic and prosperity, container carriers have to confront oil prices and climate. It gets high risk and uncertainty in operation. The threat of oversupply of capacity will affect the operation and development of container carriers. Costs is aggravated on container carriers due to they brought too many new vessels when previous boom but economic declined soon. Fleet deployment is a big issue when making operational strategies and plans. Container carriers have to plan how to purchase and lease vessels. People said that the proportion of owned container vessels of Europe container carriers is lower than Asia container carriers. Europe container carries may not spend too much cost to purchase container vessels so it would be more flexible to lease container vessels and able to deploy container vessels in/out of route and create brand new service more mobility. For this reason, operating performance is better than Asia container carriers. This study is base on secondary data from Alphaliner, Containerisation International and American Shipper database and periodical in 2007 and 2008: net profit, total revenue, the numbers of owned, leased and operated container vessels. The 15 major container carriers are divided into three segments: Asia-Chinese, Asia-Non-Chinese and Non-Asia container carriers. The General Linear Mode is practiced on these to find out if any impact or correlation from the proportion of owned container vessel to performance. The conclusions of this study are as following. 1. It¡¦s no correlation between the proportion of owned container vessels and net profit and total revenue in 2007 and 2008 2. The proportion of owned container vessels can not be fixed at a value 3. The proportion of owned container vessels is dynamic, may not be operated by container carriers.

none

Coastal dolphins are often exposed to habitat degradation and direct interactions with humans. Major factors that influence dolphin distribution, critical for conservation concerns, are still poorly understood even for the bottlenose dolphin, the best-studied cetacean. To establish the environmental conditions that best predict occurrence of bottlenose dolphins and high feeding densities in Galveston Bay, I conducted a total of 367 boat surveys in five locations of the estuary, totaling 3,814.77 km of search effort. I counted groups of dolphins and measured surface water temperature, salinity, turbidity, total number of boats, shrimp vessels, and number of seabirds. Using geospatial tools, these data -- along with location, distance to the Gulf of Mexico and water depth -- were analyzed on a 500- m resolution grid. Temporal factors at daily and seasonal scales were also examined. Occurrence was modeled using a Generalized Additive Model and core feeding densities (i.e., feeding densities above 2 SD of the mean) were modeled with a Generalized Linear Model. A total of 1,802 dolphins in 262 groups were detected, 56.87% of which were feeding. I found that all factors except warm/cold seasons and turbidity were useful to predict dolphin distribution, which was related non-linearly to most predictors. Fewer variables were relevant in predicting core feeding densities. These were, in decreasing order of relevance, distance to the Gulf of Mexico, surface water temperature, depth, number of boats, and warm/cold seasons. Feeding was highly clustered and the main core areas, less than approximately 3 km2 wide, were stable across time of day and season. The majority of the occurrences (86.2%) and feeding groups (94%) were situated in two bay locations - Galveston Ship Channel (GSC) and Bolivar Roads (BRD) - that amounted to only one- fifth of the surveyed area. Compared to conditions in GSC and BRD when no dolphins were sighted, feeding cores weredeeper and more seabirds were observed. This fine-scale study of bottlenose dolphin distribution may contribute to a better understanding of habitat requirements for coastal dolphins. It also may provide information needed to minimize potential negative impacts to this population caused by human activities.

The objective of this thesis is to examine random effect models applied to binary data. I will use classical and Bayesian inference to fit generalized linear mixed models to a specific data set. The data analyzed in this thesis comes from a study examining the injection practices of needle exchange clientele in Victoria, B.C. focusing on their risk networks. First, I will examine the application of social network analysis to the study of injection drug use, focusing on issues of gender, norms, and the problem of hidden populations. Next the focus will be on random effect models, where I will provide some background and a few examples pertaining to generalized linear mixed models (GLMMs). After GLMMs, I will discuss the nature of the injection drug use study and the data which will then be analyzed using a GLMM. Lastly, I will provide a discussion about my results of the GLMM analysis along with a summary of the injection practices of the needle exchange clientele. / Graduate / 0463