GEOTRUST SSL CERTIFICATE
Titre : | ON THE CONSISTENCY AND THE ASYMPTOTIC NORMALITY OF SEVERAL CONDITIONAL MODELS, DEPENDENT CASE |
Auteurs : | Saâdia, Rahmani, Directeur de thèse ; Seba Djillali, Auteur |
Type de document : | texte imprimé |
Editeur : | université Dr mouley tahar, Faculté des Sciences, Saida, Algerie : Alger: univ-saida, 2019 |
Format : | 67p / 27cm |
Langues: | Français |
Catégories : | |
Mots-clés: | Functional Conditional ; Application |
Résumé : |
There is actually an increasing number of estimation coming from different fields
of applied sciences, in which the collected data are curves, indeed the progress of the computing tools both in term of memory and computational capacities, allows us to deal with large sets of data. Since the middle of nineties, the different situations when functional variables can be observes has motivated the statistical development that we called "statistics for functional data". Traditional statistical methods fail as soon as we deal with the functional data, if for instance we consider a sample of finely discretized curves two crucial statistical problems appear: the first comes from the relation between the size of the sample and the number of variables, the second is due to the existence of the strong corre- lations between variables and becomes an ill-conditional problem in the context of multivariate linear model, so there is a real necessity to develop statistical models. A well-known statistical problem consists in studying the link between two variables in order to predict one of them, this problem has been widely studied for real or multivariate variables, but it also obviously occurs with functional variables. There are several ways to approach the prediction setting, and one of the most pop- ular is certainly the regression method which is based on conditional expectation. For robustness purposes we have two alternative techniques : conditional quantile and conditional mode. The disadvantage of classical regression is that the estimation of the regression function is sensitive to outliers and may be inappropriate in some cases when the distribution is multimodal or strongly asymmetrical, the problem of robustness can be solved by the prediction using conditional mode. The conditional quantile which can reveal an entire distributional relationship be- tween the covariates and the response variable is another alternative predictor to classical regression. Moreover, conditional quantiles are well-known for their ro- bustness with respect to heavy-tailed error distributions and outliers which allows to consider them as a useful alternative to the regression function. For the above reasons, conditional quantiles are used in many areas of applied research and are frequently used in a regression setup, called quantile regression. |
Note de contenu : |
1 General introduction
2 Functional Conditional Mode 3 Functional Conditional Quantile 4 Application and conclusion Bibliography |
Exemplaires (1)
Code-barres | Cote | Support | Localisation | Section | Disponibilité |
---|---|---|---|---|---|
SCT01657 | TMMS00363 | Périodique | Salle des Thèses | Mathématique | Exclu du prêt |
Documents numériques (1)
djil-seba.pdf Adobe Acrobat PDF |