Semiparametric modeling is a hybrid of the parametric and nonparametric modelings where some function forms are known and others are unknown. In this dissertation, we have made several contributions to semiparametric modeling based on the single index model related to the following three topics: the first is to propose a model for detecting change points simultaneously with estimating the unknown function; the second is to develop two models for spatially correlated data; and the third is to further develop two models for spatially-temporally correlated data. To address the first topic, we propose a unified approach in its ability to simultaneously estimate the nonlinear relationship and change points. We propose a single index change point model as our unified approach by adjusting for several other covariates. We nonparametrically estimate the unknown function using kernel smoothing and also provide a permutation based testing procedure to detect multiple change points. We show the asymptotic properties of the permutation testing based procedure. The advantage of our approach is demonstrated using the mortality data of Seoul, Korea from January, 2000 to December, 2007. On the second topic, we propose two semiparametric single index models for spatially correlated data. One additively separates the nonparametric function and spatially correlated random effects, while the other does not separate the nonparametric function and spatially correlated random effects. We estimate these two models using two algorithms based on Markov Chain Expectation Maximization algorithm. Our approaches are compared using simulations, suggesting that the semiparametric single index nonadditive model provides more accurate estimates of spatial correlation. The advantage of our approach is demonstrated using the mortality data of six cities, Korea from January, 2000 to December, 2007. The third topic involves proposing two semiparametric single index models for spatially and temporally correlated data. Our first model has the nonparametric function which can separate from spatially and temporally correlated random effects. We refer it to "semiparametric spatio-temporal separable single index model (SSTS-SIM)", while the second model does not separate the nonparametric function from spatially correlated random effects but separates the time random effects. We refer our second model to "semiparametric nonseparable single index model (SSTN-SIM)". Two algorithms based on Markov Chain Expectation Maximization algorithm are introduced to simultaneously estimate parameters, spatial effects, and times effects. The proposed models are then applied to the mortality data of six major cities in Korea. Our results suggest that SSTN-SIM is more flexible than SSTS-SIM because it can estimate various nonparametric functions while SSTS-SIM enforces the similar nonparametric curves. SSTN-SIM also provides better estimation and prediction. / Ph. D.

Ce travail de thèse étudie l'évolution des éclosions d'Aedes (Ochlerotatus) caspius (Pallas, 1771) (Culicidae) sur le littoral méditerranéen français dans un contexte de changement climatique. Aedes caspius est un moustique nuisant se développant dans les milieux naturels salés à submersions temporaires dont le cycle de vie est réglé par un nombre de paramètres environnementaux relativement limités. Nous avons tenté de mieux cerner les déterminants des dynamiques temporelles actuelle et future des éclosions d'Aedes caspius. L'étude de l'aire de répartition d'Aedes caspius à l'échelle du bassin méditerranéen a conduit à définir une enveloppe bioclimatique actuelle et future. L'ensemble des anomalies climatiques telles que envisagées par les scenarii A2 et B2 du GIEC (horizons 2020, 2050 et 2080) entraîne un élargissement de l'enveloppe bioclimatique vers le nord et l'ouest de la France, et potentiellement une extension de la zone d'intervention de l'Entente Interdépartementale pour la Démoustication Méditerranée (EID), sans toutefois exclure les zones humides actuellement démoustiquées. La dynamique temporelle a donc pu être étudiée à partir de la base de données des interventions quotidiennes de l'EID, sur la période 2004-2009. A cette échelle locale (3 105 ha) et tenant compte de la variabilité météorologique inter- et intra-annuelle sur la période, un modèle logistique binaire d'occurrences d'éclosion a été développé. / This thesis studies the evolution of hatching of Aedes (Ochlerotatus) caspius (Pallas, 1771) (Culicidae) on the French Mediterranean coast in the context of climate change. Aedes caspius is a nuisance mosquito in natural salt to temporary submersion wetlands whose life cycle is regulated by a number of environmental parameters relatively limited. We tried to understand the determinants of temporal dynamics of current and future hatching of Aedes caspius. The study of the distribution area of Aedes caspius across the Mediterranean basin has led to the definition of present and future bioclimatic envelopes. The set of climatic anomalies such as envisaged by the scenarios A2 and B2 of the Intergovernmental Panel on Climate Change (IPCC) (period 2020, 2050 and 2080) leads to an enlargement of the bioclimatic envelope to the north and west of France, and the potential extension of the present day intervention area of the Entente Interdépartementale pour la Démoustication Méditerranée (Interdepartmental Agreement for Mosquito Control Mediterranean (EID)). The temporal dynamics has been studied from the database of daily interventions of EID, over the period 2004-2009. At this local scale (3105 ha) and taking into account inter-and intra-annual meteorological variability, a binary logistic model of occurrences of daily outbreaks has been developed. The resulting model reports that the type of land use, minimum temperature, photoperiod, temperature amplitude and to a lesser extent precipitation and their variability are the main factors explaining the presence of Aedes caspius outbreaks.

The aim of this thesis is to describe calculation methods for variability esti- mation of claims reserve in non-life insurance. The thesis focuses on three main categories of models: Mack's stochastic Chain-Ladder, generalized linear models and bootstrap. Both the theoretical and also the empirical parts are included. Empirical part is devoted to application of all the models described above on both real and simulated data. 1

Neste trabalho dois novos modelos estatísticos de regressão são propostos, com estrutura muito semelhante aos Modelos Lineares Generalizados (MLG) porém, admitindo as distribuições Weibull exponenciada (WE) e beta Weibull (BW) para o componente aleatório as quais não pertencem a família exponencial como é requerido em MLG. Os novos modelos trazem uma nova abordagem para as distribuições admitidas em modelos de regressão e estende o MLG para além da família exponencial. Os modelos, nomeados por Modelo Linear Weibull Exponeciada Generalizado (MLWEG) e Modelo Linear Beta Weibull Generalizado (MLBWG), possuem como caso particular o modelo Exponencial, pertencente a família de MLG, além de outros modelos que os MLGs não contemplam como, por exemplo: Weibull, WE, Exponencial Exponenciado (EE) entre outros. Além da função taxa de falha (ftf) constante da distribuição Exponencial, os novos modelos ajustam também formas monótonas e não monótonas da ftf. Quando se admite função de ligação logarítmica obtém-se o mesmo modelo de locação e escala, muito utilizado em análise de sobrevivência, sem a necessidade de transformação da variável resposta simplificando a modelagem e permitindo maior compreensão da influência das covariáveis na resposta. Método de estudo de observações influentes foi construído baseado na metodologia de influência local sobre três esquemas de perturbações: perturbação da verossimilhança, da variável resposta e das covariáveis e a análise de resíduo foi proposta a partir da função quantílica. Por fim, dois conjuntos de dados reais foram utilizados para ilustrar a aplicabilidade dos modelos propostos e seus resultados discutidos. / In this work two new statistical regression models are proposed, with very similar structure to Generalized Linear Models (GLM) but, assuming the exponentiated Weibull (EW) and beta Weibull (BW) distributions for the random component which do not belong to the exponential family as required in GLM. The new models bring a new approach to the distribution accepted in regression models and extend the GLM beyond of the exponential family. The models, named by Generalized Exponentiated Weibull Linear Model (GEWLM) and Generalized Beta Weibull Linear Model (GBWLM) have as a particular case the Exponential model, belonging to the family of GLM, and other models that GLMs do not include, for example : Weibull, EW, Exponentiated Exponential (EE) among others. Besides the failure rate function (frf) constant of Exponential distribution, the new models also model monotonous and not monotonous forms of frf. When it accepts logarithmic link function obtains the same location and scale model, widely used in the analysis of survival without the need to transform the response variable simplifying the modeling and allowing greater understanding of the inuence of covariates on the response. Study of inuential observations method was built based on the methodology of the local inuence on three perturbations schemes: perturbation of the likelihood of the response variable and the covariates and residual analysis was proposed from the quantile function. Finally, two sets of real data are used to illustrate the applicability of the models proposed and results discussed.

Neste trabalho dois novos modelos estatísticos de regressão são propostos, com estrutura muito semelhante aos Modelos Lineares Generalizados (MLG) porém, admitindo as distribuições Weibull exponenciada (WE) e beta Weibull (BW) para o componente aleatório as quais não pertencem a família exponencial como é requerido em MLG. Os novos modelos trazem uma nova abordagem para as distribuições admitidas em modelos de regressão e estende o MLG para além da família exponencial. Os modelos, nomeados por Modelo Linear Weibull Exponeciada Generalizado (MLWEG) e Modelo Linear Beta Weibull Generalizado (MLBWG), possuem como caso particular o modelo Exponencial, pertencente a família de MLG, além de outros modelos que os MLGs não contemplam como, por exemplo: Weibull, WE, Exponencial Exponenciado (EE) entre outros. Além da função taxa de falha (ftf) constante da distribuição Exponencial, os novos modelos ajustam também formas monótonas e não monótonas da ftf. Quando se admite função de ligação logarítmica obtém-se o mesmo modelo de locação e escala, muito utilizado em análise de sobrevivência, sem a necessidade de transformação da variável resposta simplificando a modelagem e permitindo maior compreensão da influência das covariáveis na resposta. Método de estudo de observações influentes foi construído baseado na metodologia de influência local sobre três esquemas de perturbações: perturbação da verossimilhança, da variável resposta e das covariáveis e a análise de resíduo foi proposta a partir da função quantílica. Por fim, dois conjuntos de dados reais foram utilizados para ilustrar a aplicabilidade dos modelos propostos e seus resultados discutidos. / In this work two new statistical regression models are proposed, with very similar structure to Generalized Linear Models (GLM) but, assuming the exponentiated Weibull (EW) and beta Weibull (BW) distributions for the random component which do not belong to the exponential family as required in GLM. The new models bring a new approach to the distribution accepted in regression models and extend the GLM beyond of the exponential family. The models, named by Generalized Exponentiated Weibull Linear Model (GEWLM) and Generalized Beta Weibull Linear Model (GBWLM) have as a particular case the Exponential model, belonging to the family of GLM, and other models that GLMs do not include, for example : Weibull, EW, Exponentiated Exponential (EE) among others. Besides the failure rate function (frf) constant of Exponential distribution, the new models also model monotonous and not monotonous forms of frf. When it accepts logarithmic link function obtains the same location and scale model, widely used in the analysis of survival without the need to transform the response variable simplifying the modeling and allowing greater understanding of the inuence of covariates on the response. Study of inuential observations method was built based on the methodology of the local inuence on three perturbations schemes: perturbation of the likelihood of the response variable and the covariates and residual analysis was proposed from the quantile function. Finally, two sets of real data are used to illustrate the applicability of the models proposed and results discussed.

Managed lanes are a form of congestion pricing that use occupancy and toll payment requirements to utilize capacity more efficiently. How socio-spatial characteristics impact users’ travel behavior toward managed lanes is the main research question of this study. This research is a case study of the conversion of a High Occupancy Vehicle (HOV) lane to a High Occupancy Toll (HOT) lane, implemented in Atlanta I-85 on 2011. To minimize the cost and maximize the size of the collected data, an innovative and cost-effective modeling framework for socioeconomic analysis of managed lanes has been developed. Instead of surveys, this research is based on the observation of one and a half million license plates, matched to household locations, collected over a two-year study period. Purchased marketing data, which include detailed household socioeconomic characteristics, supplemented the household corridor usage information derived from license plate observations. Generalized linear models have been used to link users’ travel behavior to socioeconomic attributes. Furthermore, GIS raster analysis methods have been utilized to visualize and quantify the impact of the HOV-to-HOT conversion on the corridor commutershed. At the local level, this study conducted a comprehensive socio-spatial analysis of the Atlanta I-85 HOV to HOT conversion. At the general scale, this study enhances managed lanes’ travel demand models with respect to users’ characteristics and introduces a comprehensive modeling framework for the socioeconomic analysis of managed lanes. The methods developed through this research will inform future Traffic and Revenue Studies and help to better predict the socio-spatial characteristics of the target market.

Dirichlet regression models can be used to analyze a set of variables lying in a bounded interval that sum up to a constant (e.g., proportions, rates, compositions, etc.) exhibiting skewness and heteroscedasticity, without having to transform the data. There are two parametrization for the presented model, one using the common Dirichlet distribution's alpha parameters, and a reparametrization of the alpha's to set up a mean-and-dispersion-like model. By applying appropriate link-functions, a GLM-like framework is set up that allows for the analysis of such data in a straightforward and familiar way, because interpretation is similar to multinomial logistic regression. This paper gives a brief theoretical foundation and describes the implementation as well as application (including worked examples) of Dirichlet regression methods implemented in the package DirichletReg (Maier, 2013) in the R language (R Core Team, 2013). (author's abstract) / Series: Research Report Series / Department of Statistics and Mathematics

Bien qu’il existe un large consensus de la communauté scientifique quant au rôle des indices acoustiques dans la compréhension de la parole, les mécanismes exacts permettant la transformation d’un flux acoustique continu en unités linguistiques élémentaires demeurent aujourd’hui largement méconnus. Ceci est en partie dû à l’absence d’une méthodologie efficace pour l’identification et la caractérisation des primitives auditives de la parole. Depuis les premières études de l’interface acoustico-phonétique par les Haskins Laboratories dans les années 50, différentes approches ont été proposées ; cependant, toutes sont fondamentalement limitées par l’artificialité des stimuli utilisés, les contraintes du protocole expérimental et le poids des connaissances a priori nécessaires. Le présent travail de thèse s’est intéressé { la mise en oeuvre d’une nouvelle méthode tirant parti de la situation de compréhension de parole dégradée pour mettre en évidence les indices acoustiques utilisés par l’auditeur.Dans un premier temps, nous nous sommes appuyés sur la littérature dans le domaine visuel en adaptant la méthode des Images de Classification à une tâche auditive de catégorisation de phonèmes dans le bruit. En reliant la réponse de l’auditeur { chaque essai à la configuration précise du bruit lors de cet essai, au moyen d’un Modèle Linéaire Généralisé, il est possible d’estimer le poids des différentes régions temps-fréquence dans la décision. Nous avons illustré l’efficacité de notre méthode, appelée Image de Classification Auditive, à travers deux exemples : une catégorisation /aba/-/ada/, et une catégorisation /da/-/ga/ en contexte /al/ ou /aʁ/. Notre analyse a confirmé l’implication des attaques des formants F2 et F3, déjà suggérée par de précédentes études, mais a également permis de révéler des indices inattendus. Dans un second temps, nous avons employé cette technique pour comparer les résultats de participants musiciens experts (N=19) ou dyslexiques (N=18) avec ceux de participants contrôles. Ceci nous a permis d’étudier les spécificités des stratégies d’écoute de ces différents groupes.L’ensemble des résultats suggèrent que les Images de Classification Auditives pourraient constituer une nouvelle approche, plus précise et plus naturelle, pour explorer et décrire les mécanismes { l’oeuvre au niveau de l’interface acoustico-phonétique. / There is today a broad consensus in the scientific community regarding the involvement of acoustic cues in speech perception. Up to now, however, the precise mechanisms underlying the transformation from continuous acoustic stream into discrete linguistic units remain largely undetermined. This is partly due to the lack of an effective method for identifying and characterizing the auditory primitives of speech. Since the earliest studies on the acoustic–phonetic interface by the Haskins Laboratories in the 50’s, a number of approaches have been proposed; they are nevertheless inherently limited by the non-naturalness of the stimuli used, the constraints of the experimental apparatus, and the a priori knowledge needed. The present thesis aimed at introducing a new method capitalizing on the speech-in-noise situation for revealing the acoustic cues used by the listeners.As a first step, we adapted the Classification Image technique, developed in the visual domain, to a phoneme categorization task in noise. The technique relies on a Generalized Linear Model to link each participant’s response to the specific configuration of noise, on a trial-by-trail basis, thereby estimating the perceptual weighting of the different time-frequency regions for the decision. We illustrated the effectiveness of our Auditory Classification Image method through 2 examples: a /aba/-/ada/ categorization and a /da/-/ga/ categorization in context /al/ or /aʁ/. Our analysis confirmed that the F2 and F3 onsets were crucial for the tasks, as suggested in previous studies, but also revealed unexpected cues. In a second step, we relied on this new method to compare the results of musical experts (N=19) or dyslexics participants (N=18) to those of controls. This enabled us to explore the specificities of each group’s listening strategies.All the results taken together show that the Auditory Classification Image method may be a more precise and more straightforward approach to investigate the mechanisms at work at the acoustic-phonetic interface.

This thesis summarizes latest findings about municipal solid waste (MSW) modelling. These are used to solve multivariable version of inverse prediction problem. It is not possible to solve such problem analytically, so heuristic framework using regression models and data reconciliation was developed. As a side product, models for MSW modelling using PCA (Principal Component Analysis) and LM (Linear Model) were created. These were compared with heuristic model called RF (Random Forest). Both of these models were also used for per capita MSW modelling. Theoretical parts about generalized linear models, data reconciliation and nonlinear programming are also included.

The aim of this thesis is to determine the premium of motor third party liability insurance. Thesis explores both direct approach to premium cal- culation and calculation using separately calculated claims frequency and claims severity. Emphasis is put on the use of generalized linear models whose theory is also present. Described approaches to calculation are applied on real world data, compared with each other and then final model for premium calculation is selec- ted. Premium is determined with the inclusion of IBNR reserves and expenses. Keywords: generalized linear model, claims frequency, claims severity, net pre- mium, gross premium, motor third party liability insurance, bonus-malus. 1