Xerox copy of original. Includes bibliography.

Pour modéliser un vecteur aléatoire en présence d'une co-variable, on peut d'abord faire appel à la fonction de répartition conditionnelle. En effet, cette dernière contient toute l'information ayant trait au comportement du vecteur étant donné une valeur prise par la co-variable. Il peut aussi être commode de séparer l'étude du comportement conjoint du vecteur de celle du comportement individuel de chacune de ses composantes. Pour ce faire, on utilise la copule conditionnelle, qui caractérise complètement la dépendance conditionnelle régissant les différentes associations entre les variables. Dans chacun des cas, la mise en oeuvre d'une stratégie d'estimation et d'inférence s'avère une étape essentielle à leur utilisant en pratique. Lorsqu'aucune information n'est disponible a priori quant à un choix éventuel de modèle, il devient pertinent d'opter pour des méthodes non-paramétriques. Le premier article de cette thèse, co-écrit par Jean-François Quessy et moi-même, propose une façon de ré-échantillonner des estimateurs non-paramétriques pour des distributions conditionnelles. Cet article a été publié dans la revue Statistics and Computing. En autres choses, nous y montrons comment obtenir des intervalles de confiance pour des statistiques s'écrivant en terme de la fonction de répartition conditionnelle. Le second article de cette thèse, co-écrit par Taoufik Bouezmarni, Jean-François Quessy et moi-même, s'affaire à étudier deux estimateurs non-paramétriques de la copule conditionnelles, proposés par Gijbels et coll. en présence de données sérielles. Cet article a été soumis dans la revue Statistics and Probability Letters. Nous identifions la distribution asymptotique de chacun de ces estimateurs pour des données mélangeantes. Le troisième article de cette thèse, co-écrit par Taoufik Bouezmarni, Jean-François Quessy et moi-même, propose une nouvelle façon d'étudier les relations de causalité entre deux séries chronologiques. Cet article a été soumis dans la revue Electronic Journal of Statistics. Dans cet article, nous utilisons la copule conditionnelle pour caractériser une version locale de la causalité au sens de Granger. Puis, nous proposons des mesures de causalité basées sur la copule conditionnelle. Le quatrième article de cette thèse, co-écrit par Taoufik Bouezmarni, Anouar El Ghouch et moi-même, propose une méthode qui permette d'estimer adéquatement la copule conditionnelle en présence de données incomplètes. Cet article a été soumis dans la revue Scandinavian Journal of Statistics. Les propriétés asymptotiques de l'estimateur proposé y sont aussi étudiées. Finalement, la dernière partie de cette thèse contient un travail inédit, qui porte sur la mise en oeuvre de tests statistiques permettant de déterminer si deux copules conditionnelles sont concordantes. En plus d'y présenter des résultats originaux, cette étude illustre l'utilité des techniques de ré-échantillonnage développées dans notre premier article.

Copula functions represent a methodology which can describe the dependence structure of multi-dimension random variable, and has recently become the most significant new tool to handle risk factors in finance such as Value-at Risk( VaR) which was probably the most widely used risk measure in financial institutions. In this paper, Copula and the forecast function of Garch model are well combined, and a new method Conditional-Copula-Garch is built for measure the dependence of financial data and compute the VaR of portfolios. Copula-Garch models allow for very flexible joint distribution by splitting the marginal behaviors form the dependence relation unlike the traditional approaches for the estimation VaR, such as variance-covariance, and the Monte Carlo approaches whereas demand the joint distribution to be known. This work presents an application of the Copula-Garch model in the estimation of VaR of a portfolio composed by NASDAQ and TAIEX (Taiwan stock exchanged capitalization weighted index) stock indices.

Thesis (M.F.A.)--University of Montana, 2007. / Title from title screen. Description based on contents viewed July 30, 2007. Includes bibliographical references (p. 23).

PAULINO, V. E. N. Análise de tendência da evapotranspiração e de suas variáveis climatológicas intervenientes no Estado do Ceará-BR. 2016. 57 f. Dissertação (Mestrado em Engenharia Civil: Recursos Hídricos)-Centro de Tecnologia, Universidade Federal do Ceará, Fortaleza, 2016. / Submitted by Hohana Sanders (hohanasanders@hotmail.com) on 2017-06-29T11:07:01Z No. of bitstreams: 1 2016_dis_venpaulino.pdf: 1494002 bytes, checksum: db1bc12c416313241c9c528dec0b853c (MD5) / Rejected by Marlene Sousa (mmarlene@ufc.br), reason: Colocar titulo em ingles e corrigir nome do coorientador on 2017-06-29T15:37:35Z (GMT) / Submitted by Hohana Sanders (hohanasanders@hotmail.com) on 2017-06-29T15:47:05Z No. of bitstreams: 1 2016_dis_venpaulino.pdf: 1494002 bytes, checksum: db1bc12c416313241c9c528dec0b853c (MD5) / Approved for entry into archive by Marlene Sousa (mmarlene@ufc.br) on 2017-07-20T12:01:16Z (GMT) No. of bitstreams: 1 2016_dis_venpaulino.pdf: 1494002 bytes, checksum: db1bc12c416313241c9c528dec0b853c (MD5) / Made available in DSpace on 2017-07-20T12:01:17Z (GMT). No. of bitstreams: 1 2016_dis_venpaulino.pdf: 1494002 bytes, checksum: db1bc12c416313241c9c528dec0b853c (MD5) Previous issue date: 2016-10-20 / This study analyzes annual trends in measured meteorological variables and in the reference evaporation (ETo) estimated from them, in eight climatological stations in Ceará State, Brazil. Two statistical tests for trend analysis were used - Mann-Kendall and linear regression. It is shown that maximum temperature (Tmax) presented statistically significant increasing trend in five stations, as well as the minimum temperature (Tmin), which also showed a positive trend in three stations. Wind speed (U) and sunshine hours (S) series had positive and negative trends in two stations. The relative humidity (RH) showed negative trends in two stations and positive trend in one. These irregular trend patterns in the cited variables, only directly impacted ETo in three stations. It seems that the increasing trend in ETo was probably due to a significant increase detected in maximum temperature (Tmax) and minimum air temperature (Tmin), probably not fully offset by the decrease in wind speed (U) and relative humidity (RH). A significant increase trend in ETo trends was observed in Iguatu (+) 47.45 mm/decade, Crateus (+) 37.90 mm/decade and Fortaleza (+) 5.55 mm/decade. / A Análise das variações espaciais e temporais da evapotranspiração de referência (ETo) é importante em regiões áridas e semiáridas onde os recursos hídricos são limitados. O principal objetivo deste estudo foi analisar as tendências anuais em variáveis meteorológicas medidas e na ETo calculadas a partir delas, em oito estações climatológicas do estado do Ceará, Brasil. Foram utilizados dois testes estatísticos para a análise de tendências - Mann-Kendall e regressão linear. Mostrou-se que a temperatura máxima (Tmáx) apresentou tendência significativa positiva em cinco estações, bem como a temperatura mínima (Tmín) que também apresentou tendência significativa crescente em três estações. A velocidade do vento (U) e a insolação (S) mostraram tendências crescentes e decrescentes em duas estações. A umidade relativa (RH) motrou tendências negativas em duas estações e positivas em apenas uma estação. Esses padrões de tendências irregulares nas variáveis citadas, impactaram diretamente a ETo em três estações. Supostamente a tendência de aumento na ETo foi devido a um aumento significativo detectado na Tmáx e Tmín, que provavelmente não foi totalmente compensado pela redução na velocidade do vento (U) e na umidade relativa (RH). Um aumento significativo na ETo foi observado em Iguatu (+) 47,45 mm/década, em Crateús (+) 37,90 mm/década e em Fortaleza (+) 5,55 mm/década.

Orientador: Veronica Andrea Gonzales-Lopez / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Matematica, Estatistica e Computação Cientifica / Made available in DSpace on 2018-08-11T07:35:52Z (GMT). No. of bitstreams: 1 Bianchi_MartaCristinaColozza_M.pdf: 1867659 bytes, checksum: dbaca8bf802bf1568b0603fd248dc31f (MD5) Previous issue date: 2008 / Resumo: A crescente utilização de cópulas para modelagem de dependência em dados multivariados leva ao estudo de metodologias para o ajuste de cópulas. Este estudo é recente, assim como a plena utilização da teoria de cópulas para modelagem padrão. Grande parte das metodologias existentes ainda encontra-se em fase de estudo e somente alguns métodos foram validados recentemente. Há a necessidade de mecanismos de fácil acesso a detecção de estruturas de dependência ainda escassos na literatura. Nesta dissertação, é apresentado um método gráfico para o ajuste de cópulas, adaptado do QQplot, denominado Kendall Plot. Este método torna-se mais completo que o QQplot ao se postular a adição de bandas de confiança ao gráfico Kendall Plot, que permitem tomar uma decisão em relação a uma estrutura de dependência fixa, expressa por uma cópula, a ser testada para a amostra disponível. A redução de dimensão dos dados a uma variável unidimensional denominada BIPIT, que carrega informação a respeito da estrutura de dependência dos dados, permite a utilização da adaptação do QQplot com o fim de se testar estruturas de dependência / Abstract: The growing utilization of copulas to the dependency fitting of multi-variated data leads to the study of methodologies for copulas fitting. This study is recent, such as the complete utilization of the theory of copulas to standard fitting. Many of the existing methodologies are still in studies and only some have been recently validated. There is a need for easy-access mechanisms to detect dependency structures still missing in the statistical literature. It is presented in this dissertation a graphic method to the copulas fitting adapted from QQplot denominated Kendall Plot. This method is more complete than the QQplot due to the addition of confidence bands to the Kendall Plot graphic that allows the researcher to make a decision related to a fixed dependency structure, expressed by a copula, to be tested to the available sample. The reduction of the data dimension to a one-dimensional random variable, called BIPIT, which carries information about dependency data structure, allows the utilization of the QQplot adaptation for testing dependency structures / Mestrado / Teoria de Copulas / Mestre em Estatística

Le ranking multipartite est un problème d'apprentissage statistique qui consiste à ordonner les observations qui appartiennent à un espace de grande dimension dans le même ordre que les labels, de sorte que les observations avec le label le plus élevé apparaissent en haut de la liste. Cette thèse vise à comprendre la nature probabiliste du problème de ranking multipartite afin d'obtenir des garanties théoriques pour les algorithmes de ranking. Dans ce cadre, la sortie d'un algorithme de ranking prend la forme d'une fonction de scoring, une fonction qui envoie l'espace des observations sur la droite réelle et l'ordre finale est construit en utilisant l'ordre induit par la droite réelle. Les contributions de ce manuscrit sont les suivantes : d'abord, nous nous concentrons sur la caractérisation des solutions optimales de ranking multipartite. Le deuxième thème de recherche est la conception d'algorithmes pour produire des fonctions de scoring. Nous proposons deux méthodes, la première utilisant une procédure d'agrégation, la deuxième un schema d'approximation. Enfin, nous revenons au problème de ranking binaire afin d'établir des vitesse minimax adaptives de convergences. / Multipartite ranking is a statistical learning problem that consists in ordering observations that belong to a high dimensional feature space in the same order as the labels, so that the observations with the highest label appear at the top of the list. This work aims to understand the probabilistic nature of the multipartite ranking problem in order to obtain theoretical guarantees for ranking algorithms. In this context, the output of a ranking algorithm takes the form of a scoring function, a function that maps the space of the observation to the real line which order is induced using the values on the real line. The contributions of this manuscript are the following : First, we focus on the characterization of optimal solutions to multipartite ranking. The second research theme is the design of algorithms to produce scoring functions. We offer two methods, the first using an aggregation procedure, the second an approximation scheme. Finally, we return to the binary ranking problem to establish adaptive minimax rate of convergence.

O gênero Odontesthes é composto por 19 espécies válidas que se encontram em ambientes marinhos, estuarinos e de água doce. A primeira hipótese filogenética sugerindo o monofiletismo de Odontesthes foi realizada em 1985. Desde então, outros trabalhos, principalmente envolvendo caracteres morfológicos, corroboraram essa hipótese, porém não abrangeram a totalidade de táxons descritos para o gênero como as espécies descritas recentemente ocorrentes nas drenagens do sul do Brasil. Desse modo, o monofiletismo de Odontesthes e as relações filogenéticas entre suas espécies são aqui investigadas com base em dados morfológicos e moleculares. Além disso, com base na intensa revisão realizada, Odontesthes perugiae (espécie-tipo, conhecida por uma descrição antiga e proveniente de uma localidade incerta) é redescrita e sua distribuição é definida para as porções baixas das bacias do rio Uruguai e rio Paraná. Ainda é proposta e descrita uma nova espécie do gênero, endêmica da bacia do alto rio Uruguai, e reconhecida através de uma comparação detalhada com seus congêneres, principalmente, aqueles incluídos no “grupo perugiae”. / Odontesthes is currently composed of 19 valid species inhabiting freshwater, oceans and estuaries. The first phylogenetic hypothesis suggesting the monofiletism of Odontesthes was done in 1985. Posteriorly, other studies (mainly based on morphological characters) corroborated this hypothesis, although the southern Brazil species, recently described, one not included. In view of these facts, the phylogenetic relationships of the genus Odontesthes and its species are presented based on morphological and molecular data. In addition, Odontesthes perugiae, type species on genus, known by an old description and uncertain type-locality, is described and inhabits the lower portions of the rio Paraná and Uruguay. Also, a new species on genus endemic to the upper rio Uruguay is recognized based on a detailed comparison with its congeners, mainly, those included in the “O. perugiae species group”

Xerox copy of original. / Includes bibliography. / 3 v ; / Title page, contents and abstract only. The complete thesis in print form is available from the University Library. / Thesis (D. Litt.)--University of Adelaide, 1953

Accidental spills of hydrocarbons from underground storage tanks or pipelines are a common cause of subsurface contamination. Anthropogenic hydrocarbon contamination of soil is a global issue throughout the industrialised world. In England and Wales alone, 12% of all serious contamination incidents in 2007 were hydrocarbon related. Biodegradation could be in situ process leading to a decrease of benzene concentrations in groundwater. Recently, monitored natural attenuation has become an effective alternative to the more active remediation methods for the in situ treatment of contaminated subsurface environments. The main objective of this study was to examine the possibility of adopting monitored natural attenuation as a remediation technique for the contaminated groundwater aquifer. In this natural attenuation study, the following tasks were conducted bioremediation investigation, biological first-order decay rates, Mann-Kendall Test model and BIOSCREEN model for the contaminated groundwater aquifer. In this study, a full-scale natural bioremediation investigation was conducted at a petroleum hydrocarbon spill site. In this study, The calculated biodegradation capacity (8.261 mg/L) at this site is much higher than the detected concentrations of petroleum-hydrocarbons (3-4 mg/L) within the most contaminated area inside the plume. Thus, natural biodegradation should be able to remove the contaminants effectively. The calculated biological first-order decay rates for benzene were between 1.7¡Ñ10-3-9.0¡Ñ10-4 day-1 respectively. Mann-Kendall test was applied to analyze the trend of contaminant variations. Results show that the S-value of monitor wells SW-1W, SW-4W, SW-42W, SW-23W, SW-30W, SW-67W and SW-70W were -2.23607, -1.16276, -1.52053, -1.34164, -1.26323, 0 and -1.34164, respectively. The negative S values reveal that the all contaminants tended to decrease. This indicates that the hydrocarbon plume at this site is not expanding, and has been contained effectively by the natural attenuation mechanisms. BIOSCREEN model from the groundwater analyses indicate, a first-order decay model reached the downgradient monitor well located 220 m from the spill location. that approximately 89% of the contaminate removal was due to biodegradation processes. The study of petroleum-hydrocarbons bacterial consortium were include Aquincola tertiaricarbonis L10¡BBosea sp. GR060219¡BBrachymonas petroleovorans strain CHX¡BHydrogenophaga sp. p3(2011)¡BHydrogenophaga sp.¡BMethylibium sp. YIM 61602¡BMycobacterium sp.¡BRhodoferax sp. IMCC1723¡BRhodoferax sp.¡BUncultured Rhodocyclaceae bacterium clone Elev_16S_975¡BUncultured Rhodocyclaceae bacterium clone eub62B1¤ÎUncultured Beggiatoa sp. clone GE7GXPU01BJTWR. Thus, the in situ bioremediation technology has the potential to be developed into an environmentally, economically and naturally acceptable remediation technology. Evidences for the occurrence of natural attenuation include the following: (1) depletion of dissolved oxygen, nitrate, and sulfate; (2) production of dissolved ferrous iron, sulfide, and CO2; (3) decreased BTEX concentrations and BTEX as carbon to TOC ratio along the transport path; (4) increased alkalinity and microbial species; (5) limited spreading of the BTEX plume; and (6) preferential removal of certain BTEX components along the transport path. Results indicate that natural attenuation can effectively contain the plume, and biodegradation processes played an important role on contaminant removal.