Dissertations / Theses on the topic 'Estimateurs non-paramétriques'
Create a spot-on reference in APA, MLA, Chicago, Harvard, and other styles
Consult the top 15 dissertations / theses for your research on the topic 'Estimateurs non-paramétriques.'
Next to every source in the list of references, there is an 'Add to bibliography' button. Press on it, and we will generate automatically the bibliographic reference to the chosen work in the citation style you need: APA, MLA, Harvard, Chicago, Vancouver, etc.
You can also download the full text of the academic publication as pdf and read online its abstract whenever available in the metadata.
Browse dissertations / theses on a wide variety of disciplines and organise your bibliography correctly.
Paillisse, Pinçon Claire. "Estimateurs non-paramétriques et semi-paramétriques efficaces dans l'analyse des données censurées mutlivariées." Paris 11, 2003. http://www.theses.fr/2003PA11T047.
Full textMaillot, Bertrand. "Propriétés asymptotiques de quelques estimateurs non-paramétriques pour des variables vectorielles et fonctionnelles." Paris 6, 2008. http://www.theses.fr/2008PA066477.
Full textCamirand, Lemyre Félix. "Sur des estimateurs et des tests non-paramétriques pour des distributions et copules conditionnelles." Thèse, Université de Sherbrooke, 2016. http://hdl.handle.net/11143/9517.
Full textLhéritier, Hugo. "Comportement asymptotique de certains estimateurs sur des modèles paramétriques et sous des conditions non standard." Orléans, 2003. http://www.theses.fr/2003ORLE2005.
Full textRivoira, Arnaud. "Analyse spectrale des signaux stochastiques à échantillonage aléatoire." Paris 11, 2003. http://www.theses.fr/2003PA112157.
Full textThe work presented here deals with the spectral analysis of randomly sampled stochastic signals. After some recalls on the technical and scientific issues at stake, a state of the art of the previous methods is given and the mathematical framework is introduced. Spectral analysis methods can be classified into two categories according to whether or not they use the sampling times are used. The methods of the latter category are considered first. Following an overview of these methods, a new parametric approach, based on the identification to the CARMA model, is detailed. Then, the methods using the values of the sampling times are studied. In particular, two classes of estimators are proposed: the estimators, called IRINCORREL, which are related to those introduced by Masry, the estimators by projection, which generalize the very famous Slotting technique and its different versions. Finally, we conclude by giving a synthetic summary exhibiting the different prospects of this study and the possible extensions that could be investigated
Amiri, Aboubacar. "Estimateurs fonctionnels récursifs et leurs applications à la prévision." Phd thesis, Université d'Avignon, 2010. http://tel.archives-ouvertes.fr/tel-00565221.
Full textDebbarh, Mohammed. "Quelques propriétés asymptotiques dans les modèles additifs de régression." Paris 6, 2006. http://www.theses.fr/2006PA066020.
Full textGneyou, Kossi Essona. "Inférence statistique non paramétrique pour l'analyse du taux de panne en fiabilité : Théorèmes limites fonctionnels pour les processus produit-limite et les estimateurs non paramétriques du taux de panne dans les modèles de variables aléatoires arbitrairement censurées." Paris 6, 1991. http://www.theses.fr/1991PA066504.
Full textEl, Waled Khalil. "Estimations paramétriques et non-paramétriques pour des modèles de diffusions périodiques." Thesis, Rennes 2, 2015. http://www.theses.fr/2015REN20042/document.
Full textIn this thesis, we consider a drift estimation problem of a certain class of stochastic periodic processes when the length of observation goes to infinity. Firstly, we deal with the linear periodic signal plus noise model dζt = f (t, θ)dt + σ(t)dWt, ;and we study the parametric estimation from a continuous and discrete observation of the process f_tg throughout the interval [0; T]. Using the maximum likelihood method we show the existence of an estimator θˆT which is consistent, asymptotically normal and asymptotically efficient in the sens minimax. When f(t; _) = _f(t), the expression of ^_T is explicit and we obtain the mean square convergence in the both continuous and discrete observation cases. In addition, we deduce the strong consistency in the case of continuous observation.Secondly, we consider the nonparametric estimation problem of the function f(_) for the next two periodic models of type signal plus noise and Ornstein-Uhlenbeckd_t = f(t)dt + _(t)dWt; d_t = f(t)_tdt + dWt:For the signal plus noise model, we build a kernel estimator, the convergence in mean square uniformly over [0; P] and almost sure convergence are established, as well as the asymptotic normality. For the Ornstein-Uhlenbeck model, we prove the convergence uniformly over [0; P] of the bias and the mean square convergence. Moreover, we study the speed of these convergences
Tadj, Amel. "Sur les modèles non paramétriques conditionnels en statistique fonctionnelle." Toulouse 3, 2011. http://thesesups.ups-tlse.fr/1219/.
Full textIn this thesis, we consider the problem of the nonparametric estimation in the conditional models when the regressor takes its values in infinite dimension space. More precisely, we treated two cases when the response variable is real and functional. One establishes almost complete uniform convergence of nonparametric estimators for certain conditional models. Firstly, we consider a sequence of i. I. D. Observations. In this context, we build kernel estimators of the conditional cumulative distribution, the conditional density, the conditional hazard function and the conditional mode. We give the uniform consistency rate of these estimators. We illustrate our results by giving an application on simulated samples. Secondly, we generalize our results when the response variable is in a Banach space. We estimate the regression function. In this context, we treat both cases : i. I. D and dependent observations. In the last case, we consider that the observations are Béta-mixing and we establishes almost complete pointwise convergence. Our asymptotic results exploit the topological structure of functional space for the observations. Let us note that all the rates of convergence are based on an hypothesis of concentration of the measure of probability of the functional variable on the small balls and also on the Kolmogorov’s entropy which measures the number of the balls necessary to cover some set. Moreover, when the response variable is functional the rate of convergence contains a new term which depends on type of Banach space
Ouadah, Sarah. "Lois limites fonctionnelles pour le processus empirique et applications." Phd thesis, Université Pierre et Marie Curie - Paris VI, 2012. http://tel.archives-ouvertes.fr/tel-00766805.
Full textMagalhães, Nelo. "Validation croisée et pénalisation pour l'estimation de densité." Thesis, Paris 11, 2015. http://www.theses.fr/2015PA112100/document.
Full textThis thesis takes place in the density estimation setting from a nonparametric and nonasymptotic point of view. It concerns the statistical algorithm selection problem which generalizes, among others, the problem of model and bandwidth selection. We study classical procedures, such as penalization or resampling procedures (in particular V-fold cross-validation), which evaluate an algorithm by estimating its risk. We provide, thanks to concentration inequalities, an optimal penalty for selecting a linear estimator and we prove oracle inequalities and adaptative properties for resampling procedures. Moreover, new resampling procedure, based on estimator comparison by the mean of robust tests, is introduced as an alternative to procedures relying on the unbiased risk estimation principle. A second goal of this work is to compare these procedures from a theoretical point of view and to understand the role of V for V-fold penalization. We validate these theoretical results on empirical studies
Sow, Mohamedou. "Développement de modèles non paramétriques et robustes : application à l’analyse du comportement de bivalves et à l’analyse de liaison génétique." Thesis, Bordeaux 1, 2011. http://www.theses.fr/2011BOR14257/document.
Full textThe development of robust and nonparametric approaches for the analysis and statistical treatment of high-dimensional data sets exhibiting high variability, as seen in the environmental and genetic fields, is instrumental. Here, we model complex biological data with application to the analysis of bivalves’ behavior and to linkage analysis. The application of mathematics to the analysis of mollusk bivalves’behavior gave us the possibility to quantify and translate mathematically the animals’behavior in situ, in close or far field. We proposed a nonparametric regression model and compared three nonparametric estimators (recursive or not) of the regressionfunction to optimize the best estimator. We then characterized the biological rhythms, formalized the states of opening, proposed methods able to discriminate the behaviors, used shot-noise analysis to characterize various opening/closing transitory states and developed an original approach for measuring online growth.In genetics, we proposed a more general framework of robust statistics for linkage analysis. We developed estimators robust to distribution assumptions and the presence of outlier observations. We also used a statistical approach where the dependence between random variables is specified through copula theory. Our main results showed the practical interest of these estimators on real data for QTL and eQTL analysis
Blanchet, Thomas. "Essays on the Distribution of Income and Wealth : Methods, Estimates and Theory." Thesis, Paris, EHESS, 2020. http://www.theses.fr/2020EHES0004.
Full textThis thesis covers several topics on the distribution of income and wealth. In the first chapter, we develop a new methodology to exploit tabulations of income and wealth such as the one published by tax authorities. In it, we define generalized Pareto curves as the curve of inverted Pareto coefficients b(p), where b(p) is the ratio between average income or wealth above rank p and the p-th quantile Q(p) (i.e. b(p)=E[X|X>Q(p)]/Q(p)). We use them to characterize entire distributions, including places like the top where power laws are a good description, and places further down where they are not. We develop a method to flexibly recover the entire distribution based on tabulated income or wealth data which produces smooth and realistic shapes of generalized Pareto curves.In the second chapter, we present a new approach to combine survey data with tax tabulations to correct for the underrepresentation of the rich at the top. It endogenously determines a "merging point'' between the datasets before modifying weights along the entire distribution and replacing new observations beyond the survey's original support. We provide simulations of the method and applications to real data. The former demonstrate that our method improves the accuracy and precision of distributional estimates, even under extreme assumptions, and in comparison to other survey correction methods using external data. The empirical applications show that not only can income inequality levels change, but also trends.In the third chapter, we estimate the distribution of national income in thirty-eight European countries between 1980 and 2017 by combining surveys, tax data and national accounts. We develop a unified methodology combining machine learning, nonlinear survey calibration and extreme value theory in order to produce estimates of pre-tax and post-tax income inequality, comparable across countries and consistent with macroeconomic growth rates. We find that inequality has increased in a majority of European countries, especially between 1980 and 2000. The European top 1% grew more than two times faster than the bottom 50% and captured 18% of regional income growth.In the fourth chapter, I decompose the dynamics of the wealth distribution using a simple dynamic stochastic model that separates the effects of consumption, labor income, rates of return, growth, demographics and inheritance. Based on two results of stochastic calculus, I show that this model is nonparametrically identified and can be estimated using only repeated cross-sections of the data. I estimate it using distributional national accounts for the United States since 1962. I find that, out of the 15pp. increase in the top 1% wealth share observed since 1980, about 7pp. can be attributed to rising labor income inequality, 6pp. to rising returns on wealth (mostly in the form of capital gains), and 2pp. to lower growth. Under current parameters, the top 1% wealth share would reach its steady-state value of roughly 45% by the 2040s, a level similar to that of the beginning of the 20th century. I then use the model to analyze the effect of progressive wealth taxation at the top of the distribution
Lopez, Olivier. "Réduction de dimension en présence de données censurées." Phd thesis, Rennes 1, 2007. http://tel.archives-ouvertes.fr/tel-00195261.
Full textvariable explicative. Nous développons une nouvelle approche de réduction de la dimension afin de résoudre ce problème.