Recently, variable selection in high-dimensional data has attracted much research interest. Classical stepwise subset selection methods are widely used in practice, but when the number of predictors is large these methods are difficult to implement. In these cases, modern regularization methods have become a popular choice as they perform variable selection and parameter estimation simultaneously. However, the estimation procedure becomes more difficult and challenging when the data suffer from outliers or when the assumption of normality is violated such as in the case of heavy-tailed errors. In these cases, quantile regression is the most appropriate method to use. In this thesis we combine these two classical approaches together to produce regularized quantile regression methods. Chapter 2 shows a comparative simulation study of regularized and robust regression methods when the response variable is continuous. In chapter 3, we develop a quantile regression model with a group lasso penalty for binary response data when the predictors have a grouped structure and when the data suffer from outliers. In chapter 4, we extend this method to the case of censored response variables. Numerical examples on simulated and real data are used to evaluate the performance of the proposed methods in comparisons with other existing methods.

Submitted by JOSIANE SANTOS DE OLIVEIRA (josianeso) on 2018-11-08T13:01:52Z No. of bitstreams: 1 João Eduardo Sampaio_.pdf: 5481805 bytes, checksum: ff4fd651d904071e9a30223e3222c4dc (MD5) / Made available in DSpace on 2018-11-08T13:01:52Z (GMT). No. of bitstreams: 1 João Eduardo Sampaio_.pdf: 5481805 bytes, checksum: ff4fd651d904071e9a30223e3222c4dc (MD5) Previous issue date: 2018-08-23 / Nenhuma / O Brasil precisa de um setor siderúrgico eficiente e competitivo para enfrentar a concorrência externa. A siderurgia é um ramo da metalurgia responsável pela fabricação do aço, e dentre os processos produtivos do setor, destaca-se o processo de laminação do aço, que utiliza fornos de reaquecimento. Esta pesquisa emprega a modelagem com as técnicas do Pensamento Sistêmico e da Dinâmica de Sistemas na formulação de um modelo computacional no contexto dos fornos de reaquecimento. A partir da validação do modelo, são utilizadas a Análise Envoltória de dados, para avaliar a eficiência técnica, e a regressão Tobit, para identificar variáveis estatisticamente significantes. Essas variáveis são usadas para definição dos cenários simulados. Posteriormente, as eficiências dos cenários são avaliadas por meio de estatística descritiva. Também são avaliados alvos e folgas e é testada a hipótese de igualdade da média com o teste de Welch e Post-Hoc Kruskal-Wallis. Finalmente, é realizada a análise explicativa e, com a aplicação da técnica computacional da Rede Neural Artificial, são identificadas as variáveis prevalentes da eficiência técnica do forno de reaquecimento. Tal estudo possibilita e estimula o planejamento, a gestão e a tomada de decisão a partir da análise das melhores opções. Permite, ainda, a tomada de ações com base no conhecimento prévio, contribuindo para iniciativas pontuais e focadas na competitividade. / Brazil needs an efficient and competitive steel sector to face external competition. The siderurgy is a branch of metallurgy responsible for steelmaking, and among the productive processes in the industry the steelmaking process that uses the reheating furnaces can be highlighted. This research employs the modeling with the techniques of Systemic Thinking and Systems Dynamics in the formulation of a computational model in the context of reheating furnaces. Then, using the validated model, Data Envelopment Analysis was used, evaluating the technical efficiency and the use of the Tobit regression of statistically significant variables. These variables are used to define the simulated scenarios. Subsequently, the scenarios efficiencies were evaluated by means of descriptive statistics, evaluated targets and backlash and tested the hypothesis of equality of the average with the test of Welch and Post-Hoc Kruskal-Wallis. Finally, the explanatory analysis and identified with the application of the computational technique of the Artificial Neural Network are the prevalent variables of the technical efficiency of the reheating furnace. This study enables and stimulates planning, management and decision making based on the analysis of the best options and allows the taking of actions based on previous knowledge, and thus contributes to specific initiatives focused on competitiveness.