[pt] No seu planejamento plurianual de investimentos, as organizações do setor de Exploração e Produção (EeP) estruturam alternativas de projetos de produção de petróleo e gás natural, sujeitas a diversas restrições e a incertezas técnicas e econômicas. Como não há como assegurar que os resultados dos projetos ocorram conforme o previsto, é possível que seu retorno seja inferior ao esperado, o que, dependendo da relevância, pode provocar um efeito adverso no resultado operacional e nas condições financeiras da companhia. Nesse mérito, a dissertação apresenta e aplica um modelo de programação estocástica linear inteira mista para seleção de portfólio de projetos que permita a maximização dos resultados, com restrição de risco. A aplicação considerou dados realistas do segmento de upstream de uma empresa do setor. Para representar os cenários econômicos, optou-se pela utilização da simulação de Monte Carlo do modelo Movimento Geométrico Browniano. Com o Valor Presente Líquido como retorno e Conditional Value-at-Risk representando a medida de risco, foi possível estabelecer a fronteira eficiente do risco-retorno, com a qual o decisor pode definir uma solução de portfólio, conforme sua aversão ao risco. / [en] In their multi-annual investment planning, oil and gas companies consider alternatives of production projects, subject to a variety of constraints, and technical and economic uncertainties. Considering that it is not possible to guarantee that these projects will perform as predicted, the return can be less than expected and can lead to a significant adverse effect to the operational results and to financial conditions of a given organization. Therefore, this dissertation proposes a mixed integer linear stochastic programming model for project portfolio selection that maximizes the return with risk constraint. The application considered realistic data from the upstream segment of an oil and gas company. Monte Carlo simulation of the Geometric Brownian Motion model was considered to represent the economic scenarios. Using the Net Present Value as the function and Conditional Value-at-Risk as a risk measure, it was possible to establish the efficient frontier of risk-return, which can assist the decision-maker to define the project portfolio according to their risk aversion.

The new General Data Protection Regulation (GDPR) Act will apply to all companies within the European Union after 25 May. This means stricter legal requirements for companies that in some way store personal data. The goal of this project is therefore to make it easier for companies to meet the new legal requirements. This by creating a tool that searches file systems and visually shows the user in a graphical user interface which files contain personal data. The tool uses Named entity recognition with the Linear chain conditional random field algorithm which is a type of supervised learning method in machine learning. This algorithm is used in the project to find names and addresses in files. The different models are trained with different parameters and the training is done using the stanford NER library in Java. The models are tested by a test file containing 45,000 words where the models themselves can predict all classes to the words in the file. The models are then compared with each other using the measurements of precision, recall and F-score to find the best model. The tool also uses Regular Expression to find emails, IP numbers, and social security numbers. The result of the final machine learning model shows that it does not find all names and addresses, but that can be improved by increasing exercise data. However, this is something that requires a more powerful computer than the one used in this project. An analysis of how the Swedish language is built would also need to be done to apply the most appropriate parameters for the training of the model. / Den nya lagen General data protection regulation (GDPR) började gälla för alla företag inom Europeiska unionen efter den 25 maj. Detta innebär att det blir strängare lagkrav för företag som på något sätt lagrar personuppgifter. Målet med detta projekt är därför att underlätta för företag att uppfylla de nya lagkraven. Detta genom att skapa ett verktyg som söker igenom filsystem och visuellt visar användaren i ett grafiskt användargränssnitt vilka filer som innehåller personuppgifter. Verktyget använder Named Entity Recognition med algoritmen Linear Chain Conditional Random Field som är en typ av ”supervised” learning metod inom maskininlärning. Denna algoritm används för att hitta namn och adresser i filer. De olika modellerna tränas med olika parametrar och träningen sker med hjälp av biblioteket Stanford NER i Java. Modellerna testas genom en testfil som innehåller 45 000 ord där modellerna själva får förutspå alla klasser till orden i filen. Modellerna jämförs sedan med varandra med hjälp av mätvärdena precision, recall och F-score för att hitta den bästa modellen. Verktyget använder även Regular expression för att hitta e- mails, IP-nummer och personnummer. Resultatet på den slutgiltiga maskininlärnings modellen visar att den inte hittar alla namn och adresser men att det är något som kan förbättras genom att öka träningsdata. Detta är dock något som kräver en kraftfullare dator än den som användes i detta projekt. En undersökning på hur det svenska språket är uppbyggt skulle även också behöva göras för att använda de lämpligaste parametrarna vid träningen av modellen.

[pt] Nesta dissertação será proposta uma metodologia que contabiliza o risco de inadimplência no mercado, decorrentes de débitos não pagos à câmara de comercialização de energia elétrica (CCEE) nas estratégias de contratação de geradores renováveis. As incertezas relacionadas à geração e ao preço de curto prazo são consideradas através da simulação de cenários exógenos ao modelo como habitual em otimização estocástica. A otimização robusta é empregada através de conjuntos de incerteza poliédricos a fim de modelar a inadimplência do mercado. Dessa maneira, a metodologia proposta se baseia em um modelo matemático híbrido, robusto e estocástico. De forma mais objetiva, um modelo de dois níveis é proposto com tantos problemas de segundo nível quanto o número de cenários considerados para a produção renovável. No primeiro nível, as decisões de contratação são feitas. Em seguida, para cada cenário de geração, o problema de segundo nível encontra a pior inadimplência com base na carteira de contratos encontrados pelo primeiro nível. Para resolver o problema, o modelo de dois níveis é reescrito como um problema linear equivalente de um único nível. O perfil de risco do agente é definido por meio do conhecido valor condicional em risco (conditional value-a-risk), uma medida coerente de risco. Para ilustrar a eficácia do modelo de contratação, são realizados estudos de casos com dados realistas do sistema de energia brasileiro. / [en] In this dissertation we propose a new methodology to account for the market default risk, arising from debts not paid to the market clearing house, in the renewable generators contracting strategy. Renewable generation and spot price uncertainties are considered through exogenous simulated scenarios as customary in stochastic optimization. Robust optimization with polyhedral uncertainty sets is employed to account for the market default. Thus, the proposed methodology is based on a hybrid robust and stochastic mathematical program. More objectively, a bi-level model is proposed with as many second-level problems as the number of scenarios considered for the renewable production. In the first level, contracting decisions are made. Then, for each generation scenario, a second-level problem finds the worst-case default based on the portfolio of contracts found by the first level. To solve the problem, the bi-level model is rewritten as a single-level equivalent linear problem. The agent s risk profile is defined by means of the well-known conditional value-at-risk coherent risk measure. To illustrate the effectiveness of the contracting model, case studies are performed with realistic data from the Brazilian power system.

[pt] Predição de porosidade em reservatórios de óleo e gás representa em uma tarefa crucial e desafiadora na indústria de petróleo. Neste trabalho é proposto um novo modelo não-linear para predição de porosidade que trata fácies sedimentares como variáveis ocultas ou latentes. Esse modelo, denominado Transductive Conditional Random Field Regression (TCRFR), combina com sucesso os conceitos de Markov random fields, ridge regression e aprendizado transdutivo. O modelo utiliza volumes de impedância sísmica como informação de entrada condicionada aos valores de porosidade disponíveis nos poços existentes no reservatório e realiza de forma simultânea e automática a classificação das fácies e a estimativa de porosidade em todo o volume. O método é capaz de inferir as fácies latentes através da combinação de amostras precisas de porosidade local presentes nos poços com dados de impedância sísmica ruidosos, porém disponíveis em todo o volume do reservatório. A informação precisa de porosidade é propagada no volume através de modelos probabilísticos baseados em grafos, utilizando conditional random fields. Adicionalmente, duas novas técnicas são introduzidas como etapas de pré-processamento para aplicação do método TCRFR nos casos extremos em que somente um número bastante reduzido de amostras rotuladas de porosidade encontra-se disponível em um pequeno conjunto de poços exploratórios, uma situação típica para geólogos durante a fase exploratória de uma nova área. São realizados experimentos utilizando dados de um reservatório sintético e de um reservatório real. Os resultados comprovam que o método apresenta um desempenho consideravelmente superior a outros métodos automáticos de predição em relação aos dados sintéticos e, em relação aos dados reais, um desempenho comparável ao gerado por técnicas tradicionais de geo estatística que demandam grande esforço manual por parte de especialistas. / [en] Estimating porosity in oil and gas reservoirs is a crucial and challenging task in the oil industry. A novel nonlinear model for porosity estimation is proposed, which handles sedimentary facies as latent variables. It successfully combines the concepts of conditional random fields (CRFs), transductive learning and ridge regression. The proposed Transductive Conditional Random Field Regression (TCRFR) uses seismic impedance volumes as input information, conditioned on the porosity values from the available wells in the reservoir, and simultaneously and automatically provides as output the porosity estimation and facies classification in the whole volume. The method is able to infer the latent facies states by combining the local, labeled and accurate porosity information available at well locations with the plentiful but imprecise impedance information available everywhere in the reservoir volume. That accurate information is propagated in the reservoir based on conditional random field probabilistic graphical models, greatly reducing uncertainty. In addition, two new techniques are introduced as preprocessing steps for the application of TCRFR in the extreme but realistic cases where just a scarce amount of porosity labeled samples are available in a few exploratory wells, a typical situation for geologists during the evaluation of a reservoir in the exploration phase. Both synthetic and real-world data experiments are presented to prove the usefulness of the proposed methodology, which show that it outperforms previous automatic estimation methods on synthetic data and provides a comparable result to the traditional manual labored geostatistics approach on real-world data.

Orientador: Jose Mario Martinez Perez / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Matematica, Estatistica e Computação Cientifica / Made available in DSpace on 2018-08-10T15:09:35Z (GMT). No. of bitstreams: 1 Bueno_LuisFelipeCesardaRocha_M.pdf: 1111693 bytes, checksum: 531a933822f5dcf9cacad7dea6be5f53 (MD5) Previous issue date: 2008 / Resumo: Nesta dissertacao fazemos uma exposicao sobre alguns modelos matematicos com aplicacoes em economia. Dentre os modelos estudados destacamos a versao discreta das populares medidas de risco VaR (Value at Risk ) e C-VaR (Conditional Value at Risk ). Discutimos algumas propriedades de tais medidas, e, principalmente, expomos sobre algumas ideias para otimiza-las sob uma formulação do tipo OVO (Order Value Optimization) e propomos uma nova formulação para o problema de minimizar a VaR / Abstract: In this dissertation we make a presentation on some mathematical models with applications in economics. Among the studied models we highlight a discrete version of the popular risk measures VaR (Value at Risk) and C-VaR (Conditional Value at Risk). We discuss about some properties of such measures, and, above all, expose on some ideas for optimizing the VaR and CVaR under a OVO (Order Value Optimization) formulation and propose a new formulation to the problem of minimizing the VaR / Mestrado / Otimização / Mestre em Matemática Aplicada

Le volume des données disponibles permettant de décrire l’environnement, en particulier l’atmosphère et les océans, s’est accru à un rythme exponentiel. Ces données regroupent des observations et des sorties de modèles numériques. Les observations (satellite, in situ, etc.) sont généralement précises mais sujettes à des erreurs de mesure et disponibles avec un échantillonnage spatio-temporel irrégulier qui rend leur exploitation directe difficile. L’amélioration de la compréhension des processus physiques associée à la plus grande capacité des ordinateurs ont permis des avancées importantes dans la qualité des modèles numériques. Les solutions obtenues ne sont cependant pas encore de qualité suffisante pour certaines applications et ces méthodes demeurent lourdes à mettre en œuvre. Filtrage et lissage (les méthodes d’assimilation de données séquentielles en pratique) sont développés pour abonder ces problèmes. Ils sont généralement formalisées sous la forme d’un modèle espace-état, dans lequel on distingue le modèle dynamique qui décrit l’évolution du processus physique (état), et le modèle d’observation qui décrit le lien entre le processus physique et les observations disponibles. Dans cette thèse, nous abordons trois problèmes liés à l’inférence statistique pour les modèles espace-états: reconstruction de l’état, estimation des paramètres et remplacement du modèle dynamique par un émulateur construit à partir de données. Pour le premier problème, nous introduirons tout d’abord un algorithme de lissage original qui combine les algorithmes Conditional Particle Filter (CPF) et Backward Simulation (BS). Cet algorithme CPF-BS permet une exploration efficace de l’état de la variable physique, en raffinant séquentiellement l’exploration autour des trajectoires qui respectent le mieux les contraintes du modèle dynamique et des observations. Nous montrerons sur plusieurs modèles jouets que, à temps de calcul égal, l’algorithme CPF-BS donne de meilleurs résultats que les autres CPF et l’algorithme EnKS stochastique qui est couramment utilisé dans les applications opérationnelles. Nous aborderons ensuite le problème de l’estimation des paramètres inconnus dans les modèles espace-état. L’algorithme le plus usuel en statistique pour estimer les paramètres d’un modèle espace-état est l’algorithme EM qui permet de calculer itérativement une approximation numérique des estimateurs du maximum de vraisemblance. Nous montrerons que les algorithmes EM et CPF-BS peuvent être combinés efficacement pour estimer les paramètres d’un modèle jouet. Pour certaines applications, le modèle dynamique est inconnu ou très coûteux à résoudre numériquement mais des observations ou des simulations sont disponibles. Il est alors possible de reconstruire l’état conditionnellement aux observations en utilisant des algorithmes de filtrage/lissage dans lesquels le modèle dynamique est remplacé par un émulateur statistique construit à partir des observations. Nous montrerons que les algorithmes EM et CPF-BS peuvent être adaptés dans ce cadre et permettent d’estimer de manière non-paramétrique le modèle dynamique de l’état à partir d'observations bruitées. Pour certaines applications, le modèle dynamique est inconnu ou très coûteux à résoudre numériquement mais des observations ou des simulations sont disponibles. Il est alors possible de reconstruire l’état conditionnellement aux observations en utilisant des algorithmes de filtrage/lissage dans lesquels le modèle dynamique est remplacé par un émulateur statistique construit à partir des observations. Nous montrerons que les algorithmes EM et CPF-BS peuvent être adaptés dans ce cadre et permettent d’estimer de manière non-paramétrique le modèle dynamique de l’état à partir d'observations bruitées. Enfin, les algorithmes proposés sont appliqués pour imputer les données de vent (produit par Météo France). / The amount of both observational and model-simulated data within the environmental, climate and ocean sciences has grown at an accelerating rate. Observational (e.g. satellite, in-situ...) data are generally accurate but still subject to observational errors and available with a complicated spatio-temporal sampling. Increasing computer power and understandings of physical processes have permitted to advance in models accuracy and resolution but purely model driven solutions may still not be accurate enough. Filtering and smoothing (or sequential data assimilation methods) have developed to tackle the issues. Their contexts are usually formalized under the form of a space-state model including the dynamical model which describes the evolution of the physical process (state), and the observation model which describes the link between the physical process and the available observations. In this thesis, we tackle three problems related to statistical inference for nonlinear state-space models: state reconstruction, parameter estimation and replacement of the dynamic model by an emulator constructed from data. For the first problem, we will introduce an original smoothing algorithm which combines the Conditional Particle Filter (CPF) and Backward Simulation (BS) algorithms. This CPF-BS algorithm allows for efficient exploration of the state of the physical variable, sequentially refining exploration around trajectories which best meet the constraints of the dynamic model and observations. We will show on several toy models that, at the same computation time, the CPF-BS algorithm gives better results than the other CPF algorithms and the stochastic EnKS algorithm which is commonly used in real applications. We will then discuss the problem of estimating unknown parameters in state-space models. The most common statistical algorithm for estimating the parameters of a space-state model is based on EM algorithm, which makes it possible to iteratively compute a numerical approximation of the maximum likelihood estimators. We will show that the EM and CPF-BS algorithms can be combined to effectively estimate the parameters in toy models. In some applications, the dynamical model is unknown or very expensive to solve numerically but observations or simulations are available. It is thence possible to reconstruct the state conditionally to the observations by using filtering/smoothing algorithms in which the dynamical model is replaced by a statistical emulator constructed from the observations. We will show that the EM and CPF-BS algorithms can be adapted in this framework and allow to provide non-parametric estimation of the dynamic model of the state from noisy observations. Finally the proposed algorithms are applied to impute wind data (produced by Méteo France).

Computer hardware and software laid the foundation for fundamental innovations in science, technology, economics and society. Novel application areas generate an ever-increasing demand for computation power and storage capacities. Classic CMOS-based hardware and the von Neumann architecture are approaching their limits in miniaturization, power density and communication speed. To meet future demands, researchers work on new device technologies and architecture approaches which in turn require new algorithms and a hardware/software co-design to exploit their capabilities. Since the overall system heterogeneity and complexity increases, the challenge is to build systems with these technologies that are both correct and performant by design. Formal methods in general and model checking in particular are established verification methods in hardware design, and have been successfully applied to many hardware, software and integrated hardware/software systems. In many systems, probabilistic effects arise naturally, e.g., from input patterns, production variations or the occurrence of faults. Probabilistic model checking facilitates the quantitative analysis of performance and reliability measures in stochastic models that formalize this probabilism. The interdisciplinary research project Center for Advancing Electronics Dresden, cfaed for short, aims to explore hardware and software technologies for future information processing systems. It joins the research efforts of different groups working on technologies for all system layers ranging from transistor device research over system architecture up to the application layer. The collaborations among the groups showed a demand for new formal methods and enhanced tools to assist the design and analysis of technologies at all system layers and their cross-layer integration. Addressing these needs is the goal of this thesis. This work contributes to probabilistic model checking for Markovian models with new methods to compute two essential measures in the analysis of hardware/software systems and a method to tackle the state-space explosion problem: 1) Conditional probabilities are well known in stochastic theory and statistics, but efficient methods did not exist to compute conditional expectations in Markov chains and extremal conditional probabilities in Markov decision processes. This thesis develops new polynomial-time algorithms, and it provides a mature implementation for the probabilistic model checker PRISM. 2) Relativized long-run and relativized conditional long-run averages are proposed in this work to reason about probabilities and expectations in Markov chains on the long run when zooming into sets of states or paths. Both types of long-run averages are implemented for PRISM. 3) Symmetry reduction is an effective abstraction technique to tame the state-space explosion problem. However, state-of-the-art probabilistic model checkers apply it only after building the full model and offer no support for specifying non-trivial symmetric components. This thesis fills this gap with a modeling language based on symmetric program graphs that facilitates symmetry reduction on the source level. The new language can be integrated seamlessly into the PRISM modeling language. This work contributes to the research on future hardware/software systems in cfaed with three practical studies that are enabled by the developed methods and their implementations. 1) To confirm relevance of the new methods in practice and to validate the results, the first study analyzes a well-understood synchronization protocol, a test-and-test-and-set spinlock. Beyond this confirmation, the analysis demonstrates the capability to compute properties that are hardly accessible to measurements. 2) Probabilistic write-copy/select is an alternative protocol to overcome the scalability issues of classic resource-locking mechanisms. A quantitative analysis verifies the protocol's principle of operation and evaluates the performance trade-offs to guide future implementations of the protocol. 3) The impact of a new device technology is hard to estimate since circuit-level simulations are not available in the early stages of research. This thesis proposes a formal framework to model and analyze circuit designs for novel transistor technologies. It encompasses an operational model of electrical circuits, a functional model of polarity-controllable transistor devices and algorithms for design space exploration in order to find optimal circuit designs using probabilistic model checking. A practical study assesses the model accuracy for a lab-device based on germanium nanowires and performs an automated exploration and performance analysis of the design space of a given switching function. The experiments demonstrate how the framework enables an early systematic design space exploration and performance evaluation of circuits for experimental transistor devices.:1. Introduction 1.1 Related Work 2. Preliminaries 3. Conditional Probabilities in Markovian Models 3.1 Methods for Discrete- and Continuous-Time Markov Chains 3.2 Reset Method for Markov Decision Processes 3.3 Implementation 3.4 Evaluation and Comparative Studies 3.5 Conclusion 4. Long-Run Averages in Markov Chains 4.1 Relativized Long-Run Average 4.2 Conditional State Evolution 4.3 Implementation 4.4 Conclusion 5. Language-Support for Immediate Symmetry Reduction 5.1 Probabilistic Program Graphs 5.2 Symmetric Probabilistic Program Graphs 5.3 Implementation 5.4 Conclusion 6. Practical Applications of the Developed Techniques 6.1 Test-and-Test-and-Set Spinlock: Quantitative Analysis of an Established Protocol 6.2 Probabilistic Write/Copy-Select: Quantitative Analysis as Design Guide for a Novel Protocol 6.3 Circuit Design for Future Transistor Technologies: Evaluating Polarity-Controllable Multiple-Gate FETs 7. Conclusion Bibliography Appendices A. Conditional Probabilities and Expectations A.1 Selection of Benchmark Models A.2 Additional Benchmark Results A.3 Comparison PRISM vs. Storm B. Language-Support for Immediate Symmetry Reduction B.1 Syntax of the PRISM Modeling Language B.2 Multi-Core Example C. Practical Applications of the Developed Techniques C.1 Test-and-Test-and-Set Spinlock C.2 Probabilistic Write/Copy-Select C.3 Circuit Design for Future Transistor Technologies

[pt] A formação do portfólio de uma empresa de Petróleo envolve complexas decisões devido ao ambiente de incertezas e é de extrema importância na definição do futuro estratégico da empresa. Recentemente, a otimização de um portfólio de ativos de exploração e produção de petróleo vem sendo amplamente tratada na literatura, entretanto observa-se uma escassez de trabalhos que consideram a otimização do portfólio de refino. Este trabalho tem por objetivo propor um modelo de formação de portfólio para empresas do setor de óleo e gás, que possuem atividades tanto no segmento de exploração e produção (upstream) quanto no segmento de refino (downstream), levando em conta a integração entre ambos. Assim como nos modelos tradicionais, os preços do barril de petróleo e a produtividade dos campos serão tratadas como incertezas. O modelo proposto utilizará técnicas de programação estocástica com aversão a risco, medido pelo CVaR (Conditional Value-at-Risk). A fim de validar a metodologia proposta, um estudo de caso baseado em uma empresa de óleo e gás será apresentado. A aplicação numérica indicou que o modelo que otimiza o portfólio conjunto de upstream e downstream apresenta resultado da função objetivo até 28 por cento superior ao modelo usualmente tratado na literatura que trata apenas do portfólio de upstream. / [en] The portfolio allocation of an Oil and Gas company involves complex decisions within an uncertain environment and is extremely important in defining the firm s economical and financial future behavior. Recently, the portfolio selection problem for oil exploration and production (E&P) projects has been widely treated in the literature, however, few studies consider the optimization of the combined upstream and downstream portfolio. The purpose of this work is to propose a portfolio selection model for oil and gas companies, which operates both in exploration and production (upstream) and in refining (downstream), considering the integration between them. Crude oil prices and fields performance are the main uncertainties of the problem. The proposed model makes use of risk aversion stochastic programming techniques, measured by CVaR (conditional value at risk). To validate the proposed methodology a case study based on an Oil Company will be presented. The numerical application indicates that the model considering both upstream and downstream portfolio presents objective function results 28 percent higher than the model usually used in the literature that only optimizes the upstream portfolio.

Dans cette thèse, nous étudions les propriétés asymptotiques de paramètres fonctionnels conditionnels en statistique non paramétrique, quand la variable explicative prend ses valeurs dans un espace de dimension infinie. Dans ce cadre non paramétrique, on considère les estimateurs des paramètres fonctionnels usuels, tels la loi conditionnelle, la densité de probabilité conditionnelle, ainsi que le quantile conditionnel. Le premier travail consiste à proposer un estimateur du quantile conditionnel et de prouver sa convergence uniforme sur un sous-ensemble compact. Afin de suivre la convention dans les études biomédicales, nous considérons une suite de v.a {Ti, i ≥ 1} identiquement distribuées, de densité f, censurée à droite par une suite aléatoire {Ci, i ≥ 1} supposée aussi indépendante, identiquement distribuée et indépendante de {Ti, i ≥ 1}. Notre étude porte sur des données fortement mélangeantes et X la covariable prend des valeurs dans un espace à dimension infinie.Le second travail consiste à établir la normalité asymptotique de l’estimateur à noyau du quantile conditionnel convenablement normalisé, pour des données fortement mélangeantes, et repose sur la probabilité de petites boules. Plusieurs applications à des cas particuliers ont été traitées. Enfin, nos résultats sont appliqués à des données simulées et montrent la qualité de notre estimateur. / In this thesis, we study some asymptotic properties of conditional functional parameters in nonparametric statistics setting, when the explanatory variable takes its values in infinite dimension space. In this nonparametric setting, we consider the estimators of the usual functional parameters, as the conditional law, the conditional probability density, the conditional quantile. We are essentially interested in the problem of forecasting in the nonparametric conditional models, when the data are functional random variables. Firstly, we propose an estimator of the conditional quantile and we establish its uniform strong convergence with rates over a compact subset. To follow the convention in biomedical studies, we consider an identically distributed sequence {Ti, i ≥ 1}, here density f, right censored by a random {Ci, i ≥ 1} also assumed independent identically distributed and independent of {Ti, i ≥ 1}. Our study focuses on dependent data and the covariate X takes values in an infinite space dimension. In a second step we establish the asymptotic normality of the kernel estimator of the conditional quantile, under α-mixing assumption and on the concentration properties on small balls of the probability measure of the functional regressors. Many applications in some particular cases have been also given.

[pt] O desenvolvimento de métodos simples e efetivos para estimar o valor da curva forward de energia elétrica pode permitir que participantes do mercado precifiquem adequadamente suas posições especulativas ou defensivas. Uma ferramenta como esta poderia promover maior transparência para a definição dos preços futuros permitindo que os participantes do mercado futuro possam atuar com mais segurança e trazendo com isso um necessário aumento de liquidez. Neste trabalho apresento um modelo com dois agentes representativos que administram sua exposição ao risco através de um contrato por diferenças entre o preço futuro esperado da energia elétrica na região Sudeste no Brasil e um preço de referência. Demonstra-se que este mecanismo pode abranger todos os participantes do mercado, quer sejam especuladores ou agentes envolvidos na comercialização. A função de utilidade de cada participante é modelada utilizando uma versão Generalizada da Preferência CVaR Estendida (ECP-G) e o equilíbrio nesta transação é obtido através da minimização da diferença quadrática do equivalente certo destes agentes. Os resultados obtidos são comparados às previsões de mercado feitas por especialistas para o mesmo período e demonstram aderência dentro e fora da amostra. / [en] The development of simple and effective mechanisms to estimate the value of the forward curve of power could enable market participants to better price hedging or speculative positions. This could in turn provide transparency in future price definition to all market participants and lead to more safety and liquidity in the market for electricity futures and power derivatives. This work presents a model for two market participants, a buyer and a seller of a contract for difference on the future spot price of electricity in southwest Brazil. It is shown that this model is representative of all market participants that have exposure to the future price of power. Each participant s utility function is modelled using a Generalized Extended CVaR Preference (ECP-G) and the market equilibrium is obtained through the minimization of the quadratic difference between the certainty equivalent of both agents. The results are compared with prediction of the future spot price of power made by market specialists and found to yield reasonable results when using out of sample data.