Ce travail porte sur l'estimation des indices d'autosimilarité et de stabilité d'un processus ou champ stable fractionnaire et autosimilaire ou d'un processus stable multifractionnaire.Plus précisément, soit X un processus ou un champ stable H-autosimilaire à accroissements stationnaires (H-sssi) ou un processus stable multifractionnaire. Nous observons X aux points k/n, k=0,..., n.Nos estimations sont basées sur des variations de puissances négatives beta avec -1/2<beta<0: en effet, ces variations ont une espérance et une variance.Nous obtenons des estimateurs consistants, avec les vitesses de convergence, pour plusieurs processus H-sssi alpha-stables classiques (mouvement brownien fractionnaire, mouvement stable fractionnaire linéaire, processus de Takenaka, movement de Lévy).De plus, nous obtenons la normalité asymptotique de nos estimations pour le mouvement brownien fractionnaire et le mouvement de Lévy.Ce nouveau cadre nous permet de donner une estimation pour le paramètre d'autosimilarité H sans hypothèse sur alpha et, vice versa, nous pouvons estimer l'indice stable alpha sans hypothèse sur H.En généralisant, pour le cas d'une dimension supérieure à 1, nous obtenons également des estimateurs consistants pour H et alpha. Les résutats sont illustrés par des exemples: champ de Lévy fractionnaire, champ stable fractionnaire linéaire, champ de Takenaka.Pour les processus stables multifractionnaires, nous nous concentrons sur le mouvement brownien multifractionnaire et le processus stable multifractionnaire linéaire. Dans ces deux cas, nous obtenons la consistance des estimateurs pour la fonction d'autosimilarité à un temps donné u et pour l'indice stable alpha. / This work is concerned with the estimation of the self-similarity and the stability indices of a H-self-similarity stable process (field) or a multifractional stable process.More precisely, let X be a H-sssi (self-similar stationary increments) symmetric alpha-stable process (field) or a multifractional stable process. We observe X at points k/n, k=0,...,n.Our estimates are based on beta-negative power variations with -1/2<beta<0, thanks to the existence of expectations and covariances of these variations.We get consistent estimators, with rates of convergence, for several classical H-sssi alpha-stable processes(fractional Brownian motion, well-balanced linear fractional stable motion, Takenaka's processes, Lévy motion). Moreover, we get asymptotic normality of our estimates for fractional Brownian motion and Lévy motion.This new framework allows us to give an estimator for the self-similarity parameter H without assumptions on alpha and, vice versa, we can estimate the stable index alpha without assumptions on H.Generalizing for the case of high dimensions, we also obtain consistent estimators for H and alpha. The results are illustrated with some familiar examples: Lévy fractional Brownian field, well-balanced linear fractional stable field and Takenaka random field.For multifractional stable process, we concentrate on multifractional Brownian motion and linear multifractional stable process. In these two cases, we get the consistency of the estimators for the value of self-similarity function H at a fixed time u and for the stability index alpha.

The operating motion of a ground compactor uses high power vibrations to improve mechanical properties of a compacted ground. This motion gives a good base for the vibration analysis with an aid of Signal Processing. In this thesis, the motion of a bottom plate in a compactor is of the main interest. The thesis concerns usage of two main spectral analyzing tools, Power Spectrum estimators and Power Spectral Density estimators, presenting advantages and disadvantages in the application of a vibration analysis. Moreover, an influence of two window applications, a Flattop window, and a Hanning window, is described in relation to both analyzing approaches. The results present problems that occur when a vibration with a present modulated frequency is analyzed and how a Power Spectral Density estimator arise in a more consistent estimate over analyzed vibration spectrum. What is more, an Ordinary Deflection Shapes for a simplified bottom plate model, under different motion excitations, are presented at the end of this thesis, giving a better view of the operational motion of an analyzed system.

Researches conducted by energy distribution companies on several industry segments showed that about 30%of inductionmotors are operating undersized, i.e. with less than 70%of the rated load. Under these conditions induction motors have low efficiency. Due to the characteristics of industrial processes, traditional methods of efficiency and operation analysis can not be used. In order to meet the requirements of these applications, several methods have been proposed. Most of these methods require only the measurement of voltages, currents and in some cases rotor speed. Often these methods estimate, using different techniques, the shaft torque and calculate the output power and efficiency. The purpose of this study is to develop and implement a real-time method to monitor operating conditions and efficiency of induction motors operating at steady state. The proposed method, by means of measured three-phase voltages and currents and the equivalent circuit of the machine, estimates the shaft torque and speed. Then, based on these estimations, calculates the power output and efficiency. The machine equivalent circuit is estimated by an iterative algorithm that uses only the machine nameplate data and the stator resistance. According to present tendencies a wireless system was developed were a central computer can monitor in real time the efficiency and the operational conditions of one or more motors in a industrial plant. This document presents the theoretical development of the proposed method and the results of computer simulations under different situations. Finally it presents an experimental implementation of the method where its performance is evaluated for different motors. / Estudos realizados por companhias de distribuição de energia elétrica em diversos segmentos do setor industrial demonstraram que aproximadamente 30% dos motores de indução trifásicos analisados estavam operando subdimensionados, ou seja, com uma carga inferior a 70% da carga nominal. Nestas condições motores de indução apresentam baixo rendimento. Devido `as características dos processos industriais os métodos tradicionais de análise de eficiência e operação não podem ser utilizados. Para atender aos requisitos destas aplicações foram propostos diversos métodos que necessitam apenas das tensões, correntes e em alguns casos da velocidade do rotor. De maneira geral estes métodos estimam, utilizando diferentes técnicas, o conjugado no eixo da máquina e calculam a potência de saída e a eficiência. A proposta deste trabalho é desenvolver e implementar um método de monitoramento em tempo real das condições operacionais e da eficiência de motores de indução trifásicos operando em regime permanente. O método proposto estima, utilizando valores medidos das tensões e correntes trifásicas e o circuito equivalente da maquina, o conjugado e a velocidade do eixo e, com base nestas estimativas, calcula-se a potência de saída e a eficiência. O circuito equivalente da máquina é estimado através de um algoritmo iterativo que utiliza apenas os dados de placa da máquina e a resistência do estator. Acompanhando tendências atuais, desenvolveu-se também um sistema com comunicação sem fio (wireless) onde, a partir de um computador central pode-se monitorar em tempo real as condições operacionais e a eficiência de um ou mais motores de uma planta industrial. O presente trabalho apresenta o desenvolvimento teórico do método proposto bem como o resultado de simulações computacionais sob diferentes situações. Por fim este trabalho apresenta uma implementação experimental do método onde seu desempenho é analisado em diversos motores. / Mestre em Ciências

V této práci se zabýváme odhady populačního úhrnu s využitím pomoc- ných informací. V práci je popsán obecný regresní odhad a předpoklady, za kterých je splněna asymptotická normalita tohoto odhadu. Dále jsou zde po- psány kalibrační odhady a zmínka o jejich asymptotické ekvivalenci s obec- ným regresním odhadem. Odvozené závěry aplikujeme na data z RADIO- PROJEKTu a porovnáme je s výsledky získanými společnostmi, které tento projekt realizovali. Na závěr pomocí simulací porovnáme skutečné pravdě- podobnosti pokrytí interval· spolehlivosti pro populační úhrn spočítané na základě teorie uvedené v této práci a na základě metod společností realizu- jících RADIOPROJEKT. 1

Au cours des dernières années, des changements importants dans le domaine des assurances et des finances attirent de plus en plus l’attention sur la nécessité d’élaborer un cadre normalisé pour la mesure des risques. Récemment, il y a eu un intérêt croissant de la part des experts en assurance sur l’utilisation de l’espérance conditionnelle des pertes (CTE) parce qu’elle partage des propriétés considérées comme souhaitables et applicables dans diverses situations. En particulier, il répond aux exigences d’une mesure de risque “cohérente”, selon Artzner [2]. Cette thèse représente des contributions à l’inférence statistique en développant des outils, basés sur la convergence des intégrales fonctionnelles, pour l’estimation de la CTE qui présentent un intérêt considérable pour la science actuarielle. Tout d’abord, nous développons un outil permettant l’estimation de la moyenne conditionnelle E[X|X > x], ensuite nous construisons des estimateurs de la CTE, développons la théorie asymptotique nécessaire pour ces estimateurs, puis utilisons la théorie pour construire des intervalles de confiance. Pour la première fois, l’approche de bootstrap non paramétrique est explorée dans cette thèse en développant des nouveaux résultats applicables à la valeur à risque (VaR) et à la CTE. Des études de simulation illustrent la performance de la technique de bootstrap.

Cette thèse porte sur l'estimation de fonctions à l'aide de tests dans trois cadres statistiques différents. Nous commençons par étudier le problème de l'estimation des intensités de processus de Poisson avec covariables. Nous démontrons un théorème général de sélection de modèles et en déduisons des bornes de risque non-asymptotiques sous des hypothèses variées sur la fonction à estimer. Nous estimons ensuite la densité de transition d'une chaîne de Markov homogène et proposons pour cela deux procédures. La première, basée sur la sélection d'estimateurs constants par morceaux, permet d'établir une inégalité de type oracle sous des hypothèses minimales sur la chaîne de Markov. Nous en déduisons des vitesses de convergence uniformes sur des boules d'espaces de Besov inhomogènes et montrons que l'estimateur est adaptatif par rapport à la régularité de la densité de transition. La performance de l'estimateur est aussi évalué en pratique grâce à des simulations numériques. La seconde procédure peut difficilement être implémenté en pratique mais permet d'obtenir un résultat général de sélection de modèles et d'en déduire des vitesses de convergence sous des hypothèses plus générales sur la densité de transition. Finalement, nous proposons un nouvel estimateur paramétrique d'une densité. Son risque est contrôlé sous des hypothèses pour lesquelles la méthode du maximum de vraisemblance peut ne pas fonctionner. Les simulations montrent que ces deux estimateurs sont très proches lorsque le modèle est vrai et suffisamment régulier. Il est cependant robuste, contrairement à l'estimateur du maximum de vraisemblance. / This thesis deals with the estimation of functions from tests in three statistical settings. We begin by studying the problem of estimating the intensities of Poisson processes with covariates. We prove a general model selection theorem from which we derive non-asymptotic risk bounds under various assumptions on the target function. We then propose two procedures to estimate the transition density of an homogeneous Markov chain. The first one selects an estimator among a collection of piecewise constant estimators. The selected estimator is shown to satisfy an oracle-type inequality under minimal assumptions on the Markov chain which allows us to deduce uniform rates of convergence over balls of inhomogeneous Besov spaces. Besides, the estimator is adaptive with respect to the smoothness of the transition density. We also evaluate the performance of the estimator in practice by carrying out numerical simulations. The second procedure is only of theoretical interest but yields a general model selection theorem from which we derive rates of convergence under more general assumptions on the transition density. Finally, we propose a new parametric estimator of a density. We upper-bound its risk under assumptions for which the maximum likelihood method may not work. The simulations show that these two estimators are very close when the model is true and regular enough. However, contrary to the maximum likelihood estimator, this estimator is robust.

This paper is dedicated to the empirical analysis of the pilot trainee project Stáže ve firmách, which is considered as treatment in this analysis. The main objective of the empirical analysis is estimation of average treatment effect(ATE) and average treatment effect on treated(ATET) for characteristics like socioeconomic status and wage. Counterfactual methods for policy impact evaluation like Difference in Differences Estimator(DiD), First Differences Estimator(FD) and Propensity Score Matching(PSM) are used to estimation mentioned effects. This paper contains extension of Assignment Problem that is used for people matching purposes as alternative for PSM. This way of matching provides better control over creation of couples. Resulting pairs are more similar in selected characteristics due to better control during couples creation process.

The concern of this diploma thesis is extreme value distributions. The first part formulates and proves the limit theorem for distribution of maximum. Further there are described basic properties of class of extreme value distributions. The key role of this thesis is on non-parametric estimations of extreme value index. Primarily, Hill and moment estimator are derived, for which is, based on the results of mathematical analysis, suggested an alternative choice of optimal sample fraction using a bootstrap based method. The estimators of extreme value index are compared based on simulations from proper chosen distributions, being close to distribution of given rain-fall data series. This time series is recommended a suitable estimator and suggested choice of optimal sample fraction, which belongs to the most difficult task in the area of extreme value theory.

We consider a posteriori error estimators that can be applied to anisotropic tetrahedral finite element meshes, i.e. meshes where the aspect ratio of the elements can be arbitrarily large. Two kinds of Zienkiewicz-Zhu (ZZ) type error estimators are derived which are both based on some recovered gradient. Two different, rigorous analytical approaches yield the equivalence of both ZZ error estimators to a known residual error estimator. Thus reliability and efficiency of the ZZ error estimation is obtained. Particular attention is paid to the requirements on the anisotropic mesh. The analysis is complemented and confirmed by several numerical examples.

Bayesian networks can be used to model the respiratory system. Their structure indicate how risk factors, symptoms, and diseases are related and the Conditional Probability Tables enable predictions about a patient’s need for hospitalization. Numerous structure learning algorithms exist for discerning the structure of a Bayesian network, but none can guarantee to find the perfect structure. Employing multiple algorithms can discover relationships between variables that might otherwise remain hidden when relying on a single algorithm. The Maximum Likelihood Estimator is the predominant algorithm for learning the Conditional Probability Tables. However, it faces challenges due to the data fragmentation problem, which can compromise its predictions. Failing to hospitalize patients who require specialized medical care could lead to severe consequences. Therefore, in this thesis, the use of an XGBoost model for learning is proposed as a novel and better method since it does not suffer from data fragmentation. A Bayesian network is constructed combining several structure learning algorithms, and the predictive performance of the Maximum Likelihood Estimator and XGBoost are compared. XGBoost achieved a maximum accuracy of 86.0% compared to the Maximum Likelihood Estimator, which attained an accuracy of 81.5% in predicting future patient hospitalization. In this way, the predictive performance of Bayesian networks has been enhanced. / Bayesianska nätverk kan användas för att modellera andningssystemet. Deras struktur visar hur riskfaktorer, symtom och sjukdomar är relaterade, och de villkorliga sannolikhetstabellerna möjliggör prognoser om en patients behov av sjukhusvård. Det finns många strukturlärningsalgoritmer för att urskilja strukturen i ett bayesianskt nätverk, men ingen kan garantera att hitta den perfekta strukturen. Genom att använda flera algoritmer kan man upptäcka relationer mellan variabler som annars kan förbli dolda när man bara förlitar sig på en enda algoritm. Maximum Likelihood Estimator är den dominerande algoritmen för att lära sig de villkorliga sannolikhetstabellerna. Men den står inför utmaningar på grund av datafragmenteringsproblemet, vilket kan äventyra dess prognoser. Att inte lägga in patienter som behöver specialiserad medicinsk vård kan leda till allvarliga konsekvenser. Därför föreslås i denna avhandling användningen av en XGBoost-modell för inlärning som en ny och bättre metod eftersom den inte lider av datafragmentering. Ett bayesianskt nätverk byggs genom att kombinera flera strukturlärningsalgoritmer, och den prediktiva prestandan för Maximum Likelihood Estimator och XGBoost jämförs. XGBoost uppnådde en maximal noggrannhet på 86,0% jämfört med Maximum Likelihood Estimator, som uppnådde en noggrannhet på 81,5% för att förutsäga framtida patientinläggning. På detta sätt har den prediktiva prestandan för bayesianska nätverk förbättrats.