The worldwide challenges to achieve cost effective protection against global warming impacts and to acquire reliable decision making tools continually force new developments in the area of climate change research. Climate change impacts projections involve several steps: emission scenarios generation, Global Circulation Models and Regional Climate Models (GCM/RCM) runs, downscaling, impact model running, analysis of results and decision making. Unfortunately, GCM/RCMs outputs are often biased and need to be processed before being fed into impact models. This thesis describes the effort carried out to alleviate the burden of downscaling coarse hydro-climatology data outputs from GCM/RCM and making results readily available for climate change impact analysis for specific regions, particularly in the African continent. GCM/RCM outputs are highly unreliable at the sub-grid scale to be used for region specific impact analysis (Wilby, Hay, & Leavesly, 1999). Furthermore, raw GCM/RCM outputs are often downscaled under the premises that the latter offer very coarse spatial resolution. The Internet is a common resource for users of climate change data to access relevant information. Web-based interfaces offer users the capability to retrieve such data. This thesis involves the development of a new web-portal, which addresses the demand for climate change data at the daily scale. It is a user-friendly interactive web-based interface with multiple functionalities including: capacity to process information, capacity to search, sort, retrieve and filter data and download features. Six climate variables are considered in this project: precipitation, maximum temperature, minimum temperature, wind speed, relative humidity and solar radiation. The aforementioned climate variables have been downscaled to specific geographical locations and results have been made available at a fine temporal resolution – the daily scale. The data portal currently hosts climate change data for nine stations in western Africa: Agadez, Brini N’Konni, Gaya, Maine Soroa, Maradi Airport, Niamey Airport, Tahoua, Tillabery and Zinder Airport. The above mentioned climate stations are all located in Niger. Nonetheless, the project aims to expand and cover further ground in Africa. Quantile - Quantile downscaling, also known as Quantile-Quantile mapping, matching or transformation is a statistical procedure used in this project to downscale raw GCM/RCM outputs. GCM/RCM outputs from the AMMA-Ensemble sets under the SRES A1B scenario were used as raw data.

Les applications les plus courantes des méthodes non paramétriques concernent l'estimation d'une fonction de régression (i.e. de l'espérance conditionnelle). Cependant, il est souvent intéressant de modéliser les quantiles conditionnels, en particulier lorsque la moyenne conditionnelle ne permet pas de représenter convenablement l'impact des covariables sur la variable dépendante. De plus, ils permettent d'obtenir des graphiques plus compréhensibles de la distribution conditionnelle de la variable dépendante que ceux obtenus avec la moyenne conditionnelle. A l'origine, la "quantification" était utilisée en ingénierie du signal et de l'information. Elle permet de discrétiser un signal continu en un nombre fini de quantifieurs. En mathématique, le problème de la quantification optimale consiste à trouver la meilleure approximation d'une distribution continue d'une variable aléatoire par une loi discrète avec un nombre fixé de quantifieurs. Initialement utilisée pour des signaux univariés, la méthode a été étendue au cadre multivarié et est devenue un outil pour résoudre certains problèmes en probabilités numériques.Le but de cette thèse est d'appliquer la quantification optimale en norme Lp à l'estimation des quantiles conditionnels. Différents cas sont abordés :covariable uni- ou multidimensionnelle, variable dépendante uni- ou multivariée. La convergence des estimateurs proposés est étudiée d'un point de vue théorique. Ces estimateurs ont été implémentés et un package R, nommé QuantifQuantile, a été développé. Leur comportement numérique est évalué sur des simulations et des données réelles. / One of the most common applications of nonparametric techniques has been the estimation of a regression function (i.e. a conditional mean). However it is often of interest to model conditional quantiles, particularly when it is felt that the conditional mean is not representative of the impact of the covariates on the dependent variable. Moreover, the quantile regression function provides a much more comprehensive picture of the conditional distribution of a dependent variable than the conditional mean function. Originally, the "quantization'" was used in signal and information theories since the fifties. Quantization was devoted to the discretization of a continuous signal by a finite number of "quantizers". In mathematics, the problem of optimal quantization is to find the best approximation of thecontinuous distribution of a random variable by a discrete law with a fixed number of charged points. Firstly used for a one-dimensional signal, themethod has then been developed in the multi-dimensional case and extensively used as a tool to solve problems arising in numerical probability.The goal of this thesis is to study how to apply optimal quantization in Lp-norm to conditional quantile estimation. Various cases are studied: one-dimensional or multidimensional covariate, univariate or multivariate dependent variable. The convergence of the proposed estimators is studied from a theoretical point of view. The proposed estimators were implemented and a R package, called QuantifQuantile, was developed. Numerical behavior of the estimators is evaluated through simulation studies and real data applications. / Doctorat en Sciences / info:eu-repo/semantics/nonPublished

The application area of Fifth Generation New Radio (5G-NR) called Ultra-Reliable and Low-Latency Communication (URLLC) requires a reliability, the probability of receiving and decoding a data packet correctly, of 1 - 10^5. For this requirement to be fulfilled in a resource-efficient manner, it is necessary to have a good estimation of extremely low quan- tiles of the channel quality distribution, so that appropriate resources can be distributed to users of the network system.  This study proposes and evaluates two methods for estimating extreme quantiles of the downlink channel quality distribution, linear quantile regression and Quantile Regression Neural Network (QRNN). The models were trained on data from Ericsson’s system-level radio network simulator, and evaluated on goodness of fit and resourcefulness. The focus of this study was to estimate the quantiles 10^2, 10^3 and 10^4 of the distribution.  The results show that QRNN generally performs better than linear quantile regression in terms of pseudoR2, which indicates goodness of fit, when the sample size is larger. How- ever, linear quantile regression was more effective for smaller sample sizes. Both models showed difficulty estimating the most extreme quantiles. The less extreme quantile to esti- mate, the better was the resulting pseudoR2-score. For the largest sample size, the resulting pseudoR2-scores of the QRNN was 0.20, 0.12 and 0.07, and the scores of linear quantile regression was 0.16, 0.10 and 0.07 for the respective quantiles 10^2, 10^3 and 10^4.  It was shown that both evaluated models were significantly more resourceful than us- ing the average of the 50 last measures of channel quality subtracted with a fixed back-off value as a predictor. QRNN had the most optimistic predictions. If using the QRNN, theo- retically, on average 43% more data could be transmitted while fulfilling the same reliability requirement than by using the fixed back-off value.

Yes / This study contributes to the existing literature on the relationship between oil market shocks and the green bond market by investigating the impact of the COVID-19 pandemic on their dynamic correlation. We first decompose the oil market shocks into components using a time-frequency framework. Then, we combine wavelet decomposition and quantile coherence and causality methods to discuss changes during the COVID-19 era. We observe positive effects of both supply-driven and demand-driven oil shocks on the green bond market at most quantile levels. However, supply-driven oil price changes play a major role. The results also indicate that long-term changes have a greater impact than short-term changes on the connection between oil and green bond markets. Nevertheless, the COVID-19 pandemic changed the nature of the causal relationship, as we observed no relationship under extreme market conditions during the pandemic era. We argue that the economic and social impacts of the COVID-19 pandemic have left investors focusing on the short-term substitution between oil and green bond markets. / This research was supported by the Major Projects of the National Natural Science Fund of China [NO. 71991483], the Natural Science Fund of Hunan Province [NO. 2022JJ40647] and the Fundamental Research Funds for the Central Universities of Central South University [NO. 2022ZZTS0353]. / The full-text of this article will be released for public view at the end of the publisher embargo on 06 Oct 2024.

Soient $X$ et $Y$ deux variables aléatoires de fonctions de répartition $F$ et $G$ respectivement. Deux réalisations données $x$ et $y$ sont dites équivalentes si et seulement si $F(x)=G(y)$. Cette équation est connue sous le nom ``équation équipercentile''. Sa résolution, pour un $x$ fixé, permet d'exprimer l'équivalent équipercentile de $x$ comme suit: $y(x)=G^{-1}(F(x))$, où $G^{-1}$ désigne la fonction inverse de $G$. Dans ce travail, nous proposons cinq scénarios d'estimation de la fonction d'égalisation équipercentile $G^{-1}(F(x))$. Les estimateurs proposés reposent sur l'approche de l'ajustement polynômial local. Les résultats obtenus sont les suivants. D'abord, nous montrons la convergence uniforme presque sûre des estimateurs. Ensuite, nous établissons l'approximation par un pont brownien approprié et évaluons la performance des estimateurs en utilisant l'erreur en moyenne quadratique comme mesure de perte. Finalement, nous proposons quelques simulations sous R pour illustrer nos résultats et comparons les estimateurs en les appliquant sur un jeu de données réelles provenant d'une étude longitudinale multi-centrique de la cohorte ANRS C08.

The aim of the work in this thesis is to carry out dimension reduction (DR) for high dimensional (HD) data by using statistical methods for variable selection, feature extraction and a combination of the two. In Chapter 2, the DR is carried out through robust feature extraction. Robust canonical correlation (RCCA) methods have been proposed. In the correlation matrix of canonical correlation analysis (CCA), we suggest that the Pearson correlation should be substituted by robust correlation measures in order to obtain robust correlation matrices. These matrices have been employed for producing RCCA. Moreover, the classical covariance matrix has been substituted by robust estimators for multivariate location and dispersion in order to get RCCA. In Chapter 3 and 4, the DR is carried out by combining the ideas of variable selection using regularisation methods with feature extraction, through the minimum average variance estimator (MAVE) and single index quantile regression (SIQ) methods, respectively. In particular, we extend the sparse MAVE (SMAVE) reported in (Wang and Yin, 2008) by combining the MAVE loss function with different regularisation penalties in Chapter 3. An extension of the SIQ of Wu et al. (2010) by considering different regularisation penalties is proposed in Chapter 4. In Chapter 5, the DR is done through variable selection under Bayesian framework. A flexible Bayesian framework for regularisation in quantile regression (QR) model has been proposed. This work is different from Bayesian Lasso quantile regression (BLQR), employing the asymmetric Laplace error distribution (ALD). The error distribution is assumed to be an infinite mixture of Gaussian (IMG) densities.

The U.S.-American foreclosure crisis and related economic crises have had severe and wide-reaching effects for the global economy, homeowners, and municipalities alike. These negative changes led to federal, state, regional, and local responses intended to prevent and mitigate foreclosures. As of yet, no research has examined the community- and neighborhood-level impacts of local foreclosure responses. This research seeks to determine the economic, physical, social, and political changes that resulted from these responses. A mixed methods case study of Cuyahoga County, Ohio, home to Cleveland, was used to identify local level foreclosure responses—i.e. those carried out at the county level and below—and their effects. The qualitative component was comprised of semi-structured stakeholder interviews, including local governmental representatives, advocacy groups, and neighborhood representatives. Two community subcases were investigated in depth to further examine the mechanisms and effects of foreclosure responses. The quantitative component supplements the qualitative component by means of a quantile regression model that examines relationships between foreclosure responses and changes in property value at the Census tract level, used to approximate communities. The model integrates data for the entire county and estimates coefficients at various quantiles of the dependent variable, which uncovers variations in the associations between the variables along the dependent variable’s distribution. That is, with quantile regression it is possible to determine whether foreclosure responses have different effects depending on community conditions. The results indicate that the national and local context are of particular importance when responding to the foreclosure crisis. Lackluster national level responses necessitated creative and innovative responses at the local level. The Cleveland region is characterized a weak housing market and its concomitant vacancy and abandonment problems. Thus, post-foreclosure responses that deal with blighted property are essential. A wide variety of foreclosure responses took place in Cuyahoga County, in the form of systems reform, foreclosure prevention, targeting, property acquisition and control, legal efforts, and community- and neighborhood-level efforts. Several strategies used in these responses emerged as themes: targeting, addressing blight, strengthening the social fabric, planning for the future, building institutions and organizational capacity, and advocacy. Physical and economic impacts are closely linked and are brought about especially by responses using targeting and blight reduction strategies. Social impacts, such as increased identification with, investment in, and commitment to the community occurred as the result of responses that used the strategies of strengthening the social fabric and planning a shared future for the community. Finally, the strategies of building institutions and organizational capacity and advocacy resulted in increased political power in the form of more local control and additional resources for neighborhoods and communities. These results provide deeper insight into the effects of the foreclosure crisis and local responses to it on neighborhoods and communities. This case study identifies the importance of targeting, blight removal, strengthening social bonds, planning for a shared future, increasing organizational capacity, and advocacy in addressing the foreclosure crisis on the community and neighborhood levels, especially in weak housing market cities where need far outstrips the available resources.

The purpose of this essay is to study what impact immigration has on the Swedish income distribution for the period 1992-2005. This essay uses a two-folded approach to study the income distribution, first, an income inequality measure will be investigated in order to find if the inequality increases or decreases by the increased immigration. Secondly, we estimate a quantile regression for the 10th, 50th and 90th percentiles for the period 1992, 1995, 2000 and2005, together with an OLS regression in order to find the income gap between the immigrants and natives, which is analysed for males and females separately. The study found that the inflow of immigrants increased income inequality in the lower tail of the income distribution. Immigrants at the upper tail of the income distribution are doing relatively better than the immigrants in the lower tail of the income distribution. Conclusively, independently of gender, the income gap between immigrants and natives is almost three times as large in the lower tail of the income distribution relative to the upper tail of the income distribution.

Estudar a estrutura de remuneração de um país é importante por várias razões. Talvez uma das principais seja porque a remuneração pode afetar a atratividade da carreira e a habilidade dos profissionais que exercem a função de professor. O presente trabalho buscou comparar a diferença do salário-hora entre professores e não professores graduados em carreiras tipicamente relacionadas à profissão docente: Ciências da Educação, Formação de Professores, Língua Materna, Matemática, Biologia e Química. O objetivo era entender como profissionais recém-graduados escolheram diferentes ocupações (docente versus não docente) levando ao aparecimento de um diferencial salarial importante já nos primeiros anos de vida profissional. Desta análise, sob a hipótese de ignorabilidade e a hipótese de preferências semelhantes antes do ingresso no ensino superior corretamente especificadas, pode-se concluir que aqueles que exercem a profissão de professor, tanto na média, quanto no quantil 10 e na mediana, possuem um diferencial de salário hora positivo com relação aos que não são docentes. Já no quantil 90, não há diferença de salário entre os grupos. Os resultados mostram também que a diferença é explicada majoritariamente pelo retorno às características que determinam o salário e menos por diferenças nos níveis de tais características. Existem duas implicações desses resultados. Primeiro, considerando-se salário como proxy para habilidade, pode-se inferir que os professores já são os mais habilidosos da amostra analisada. Sendo assim, aumentar o salário dos docentes não atrairia profissionais mais habilidosos para a profissão, dentro das carreiras estudadas, pois já são os profissionais mais habilidosos que são docentes. Uma alternativa seria capacitar esses profissionais de forma a aumentar a habilidade destes. Segundo, outra forma de aumentar a habilidade dos professores seria compreender como esse diferencial salarial afeta a decisão de ingresso na carreira docente, e buscar atrair profissionais mais habilidosos para o magistério antes da escolha da carreira a ser cursada no ensino superior. Isso poderia possibilitar que indivíduos com maior habilidade se interessassem pelo magistério. Em ambos os casos, para que seja efetivo para a geração de capital humano, um aumento salarial do cargo de professor deve estar atrelado à criação de condições para o aumento da habilidade dos profissionais que ocupam tal posição. / Studying the compensation structure of a country is important for several reasons. Perhaps the most important of them is because the compensation can affect the attractiveness of the teaching profession and the professional skill of performing the job. This study aimed to compare the difference in hourly wage between teachers and teachers in undergraduate careers typically related to the teaching profession: Education Sciences, Teacher Training, First Language, Mathematics, Biology and Chemistry. The objective was to understand how newly graduated professionals chose different occupations (teaching versus non-teaching) leading to the appearance of a major wage differential in the first years of working life. This analysis, under the hypothesis and the hypothesis ignorabilidade similar preferences before entering higher education properly specified, one can conclude that those who exercise the teaching profession, both on average and in 10 quantile and median of a positive differential hourly wage with respect to non-teachers. You quantile 90, there is no difference in pay between groups. The results also show that the difference is largely explained by the return of the characteristics that determine wages and less by differences in the levels of these characteristics. There are two implications of these results. First, considering salary as textit proxy for skill, it can be inferred that teachers are already the most skilled of the sample analyzed. Thus, increasing teacher salaries would attract more skilled professionals for the profession within the careers studied because they are already the most skilled professionals who are teachers. One such alternative would enable professionals to increase the ability of these. Second, another way to increase the ability of teachers would understand how this wage gap affects the decision of entering the teaching career, and seek to attract more skilled professionals for teaching before the choice of higher education. This could enable individuals with greater ability to be interested by the magisterium. In both cases, to be effective for the generation of human capital, a raise from his professorship should be linked to the creation of conditions for increasing the ability of professionals who hold such a position.

Quantile regression (QR) (Koenker and Bassett, 1978), as a comprehensive extension to standard mean regression, has been steadily promoted from both theoretical and applied aspects. Bayesian quantile regression (BQR), which deals with unknown parameter estimation and model uncertainty, is a newly proposed tool of QR. This thesis aims to make some novel contributions to the following three issues related to QR. First, whereas QR for continuous responses has received much attention in literatures, QR for discrete responses has received far less attention. Second, conventional QR methods often show that QR curves crossing lead to invalid distributions for the response. In particular, given a set of covariates, it may turn out, for example, that the predicted 95th percentile of the response is smaller than the 90th percentile for some values of the covariates. Third, mean-based clustering methods are widely developed, but need improvements to deal with clustering extreme-type, heavy tailed-type or outliers problems. This thesis focuses on methods developed over these three challenges: modelling quantile regression with discrete responses, ensuring non-crossing quantile curves for any given sample and modelling tails for collinear data with outliers. The main contributions are listed as below: * The first challenge is studied in Chapter 2, in which a general method for Bayesian inference of regression models beyond the mean with discrete responses is developed. In particular, this method is developed for both Bayesian quantile regression and Bayesian expectile regression. This method provides a direct Bayesian approach to these regression models with a simple and intuitive interpretation of the regression results. The posterior distribution under this approach is shown to not only be coherent to the response variable, irrespective of its true distribution, but also proper in relation to improper priors for unknown model parameters. * Chapter 3 investigates a new kernel-weighted likelihood smoothing quantile regression method. The likelihood is based on a normal scale-mixture representation of an asymmetric Laplace distribution (ALD). This approach benefits of the same good design adaptation just as the local quantile regression (Spokoiny et al., 2014) does and ensures non-crossing quantile curves for any given sample. * In Chapter 4, we introduce an asymmetric Laplace distribution to model the response variable using profile regression, a Bayesian non-parametric model for clustering responses and covariates simultaneously. This development allows us to model more accurately for clusters which are asymmetric and predict more accurately for extreme values of the response variable and/or outliers. In addition to the three major aforementioned challenges, this thesis also addresses other important issues such as smoothing extreme quantile curves and avoiding insensitive to heteroscedastic errors as well as outliers in the response variable. The performances of all the three developments are evaluated via both simulation studies and real data analysis.