No description available.

No description available.

Spatial analysis of species for which there is limited quantity of data, termed as the data-poor species, has been challenging due to limited information, especially lack of spatially explicit information. However, these species are frequently of high ecological, conservation and management interest. In this study, I used two empirical examples to demonstrate spatial analysis for two kinds of data-poor species. One example was seabird bycatch from the U.S. Atlantic pelagic longline fishery, which focused on rare events/species for which data are generally characterized by a high percentage of zero observations. The other example was endangered white abalone off the California coast, which focused on endangered species whose data are very limited. With the seabird bycatch example, I adopted a spatial filtering technique to incorporate spatial patterns and to improve model performance. The model modified with spatial filters showed superior performance over other candidate models. I also applied the geographically weighted approach to explore spatial nonstationarity in seabird bycatch, i.e., spatial variation in the parameters that describe relationships between biological processes and environmental factors. Estimates of parameters exhibited high spatial variation. With the white abalone example, I demonstrated the spatially explicit hierarchical demographic model and conducted a risk assessment to evaluate the efficacy of hypothetical restoration strategies. The model allowed for the Allee effect (i.e., density-dependent fertilization success) by using spatial explicit density estimates. Restoration efforts directed at larger-size individuals may be more effective in increasing population density than efforts focusing on juveniles. I also explored the spatial nonstationarity in white abalone catch data. I estimated the spatially explicit decline rate and linked the decline rate to environmental factors including water depth, distance to California coast, distance to land, sea surface temperature and chlorophyll concentration. The decline rate showed spatial variation. I did not detect any significant associations between decline rate and these five environmental factors. Through such a study, I am hoping to provide insights on applying or adapting existing methods to model spatial dynamics of data-poor species, and on utilizing information from such analyses to aid in their conservation and management. / Ph. D.

Generalized linear mixture models (GLMM) are widely used in statistical applications to model count and binary data. We consider the problem of nonparametric likelihood estimation of mixing distributions in GLMM's with multiple random effects. The log-likelihood to be maximized has the general form l(G)=Σi log∫f(yi,γ) dG(γ) where f(.,γ) is a parametric family of component densities, yi is the ith observed response dependent variable, and G is a mixing distribution function of the random effects vector γ defined on Ω. The literature presents many algorithms for maximum likelihood estimation (MLE) of G in the univariate random effect case such as the EM algorithm (Laird, 1978), the intra-simplex direction method, ISDM (Lesperance and Kalbfleish, 1992), and vertex exchange method, VEM (Bohning, 1985). In this dissertation, the constrained Newton method (CNM) in Wang (2007), which fits GLMM's with random intercepts only, is extended to fit clustered datasets with multiple random effects. Owing to the general equivalence theorem from the geometry of mixture likelihoods (see Lindsay, 1995), many NPMLE algorithms including CNM and ISDM maximize the directional derivative of the log-likelihood to add potential support points to the mixing distribution G. Our method, Direct Search Directional Derivative (DSDD), uses a directional search method to find local maxima of the multi-dimensional directional derivative function. The DSDD's performance is investigated in GLMM where f is a Bernoulli or Poisson distribution function. The algorithm is also extended to cover GLMM's with zero-inflated data. Goodness-of-fit (GOF) and selection methods for mixed models have been developed in the literature, however their application in models with nonparametric random effects distributions is vague and ad-hoc. Some popular measures such as the Deviance Information Criteria (DIC), conditional Akaike Information Criteria (cAIC) and R2 statistics are potentially useful in this context. Additionally, some cross-validation goodness-of-fit methods popular in Bayesian applications, such as the conditional predictive ordinate (CPO) and numerical posterior predictive checks, can be applied with some minor modifications to suit the non-Bayesian approach. / Graduate / 0463 / rabihsaab@gmail.com

Generalized linear mixture models (GLMM) are widely used in statistical applications to model count and binary data. We consider the problem of nonparametric likelihood estimation of mixing distributions in GLMM's with multiple random effects. The log-likelihood to be maximized has the general form l(G)=Σi log∫f(yi,γ) dG(γ) where f(.,γ) is a parametric family of component densities, yi is the ith observed response dependent variable, and G is a mixing distribution function of the random effects vector γ defined on Ω. The literature presents many algorithms for maximum likelihood estimation (MLE) of G in the univariate random effect case such as the EM algorithm (Laird, 1978), the intra-simplex direction method, ISDM (Lesperance and Kalbfleish, 1992), and vertex exchange method, VEM (Bohning, 1985). In this dissertation, the constrained Newton method (CNM) in Wang (2007), which fits GLMM's with random intercepts only, is extended to fit clustered datasets with multiple random effects. Owing to the general equivalence theorem from the geometry of mixture likelihoods (see Lindsay, 1995), many NPMLE algorithms including CNM and ISDM maximize the directional derivative of the log-likelihood to add potential support points to the mixing distribution G. Our method, Direct Search Directional Derivative (DSDD), uses a directional search method to find local maxima of the multi-dimensional directional derivative function. The DSDD's performance is investigated in GLMM where f is a Bernoulli or Poisson distribution function. The algorithm is also extended to cover GLMM's with zero-inflated data. Goodness-of-fit (GOF) and selection methods for mixed models have been developed in the literature, however their application in models with nonparametric random effects distributions is vague and ad-hoc. Some popular measures such as the Deviance Information Criteria (DIC), conditional Akaike Information Criteria (cAIC) and R2 statistics are potentially useful in this context. Additionally, some cross-validation goodness-of-fit methods popular in Bayesian applications, such as the conditional predictive ordinate (CPO) and numerical posterior predictive checks, can be applied with some minor modifications to suit the non-Bayesian approach. / Graduate / 0463 / rabihsaab@gmail.com

Savoir comment les patients choisissent les hôpitaux est d'une importance majeure non seulement pour les gestionnaires des hôpitaux mais aussi pour les décideurs. Il s'agit entre autres pour les premiers, de la gestion des flux et l'offre des soins et pour les seconds, l'implémentation des reformes dans le système de santé.Nous proposons dans cette thèse différentes modélisations des données d'admission de patients en fonction de la distance par rapport à un hôpital afin de prévoir le flux des patients et de comparer son attractivité par rapport à d'autres hôpitaux. Par exemple, nous avons utilisé des modèles bayésiens hiérarchiques pour des données de comptage avec possible dépendance spatiale. Des applications on été faites sur des données d'admission de patients dans la région de Languedoc-Roussillon.Nous avons aussi utilisé des modèles de choix discrets tels que les RUMs. Mais vu certaines limites qu'ils présentent pour notre objectif, nous avons relâché l'hypothèse de maximisation d'utilité pour une plus souple et selon laquelle un agent (patient) peut choisir un produit (donc hôpital) dès lors que l'utilité que lui procure ce produit a atteint un certain seuil de satisfaction, en considérant certains aspects. Une illustration de cette approche est faite sur trois hôpitaux de l'Hérault pour les séjours dus à l'asthme en 2009 pour calculer l'envergure territorial d'un hôpital donné . / Understanding how patients choose hospitals is of utmost importance for both hospitals administrators and healthcare decision makers; the formers for patients incoming tide and the laters for regulations.In this thesis, we present different methods of modelling patients admission data in order to forecast patients incoming tide and compare hospitals attractiveness.The two first method use counting data models with possible spatial dependancy. Illustration is done on patients admission data in Languedoc-Roussillon.The third method uses discrete choice models (RUMs). Due to some limitations of these models according to our goal, we introduce a new approach where we released the assumption of utility maximization for an utility-threshold ; that is to say that an agent (patient) can choose an alternative (hospital) since he thinks that he can obtain a certain level of satisfaction of doing so, according to some aspects. Illustration of the approach is done on 2009 asthma admission data in Hérault.