Document type
Bachelor thesisPublication date
Publication license
Please use this identifier to cite or link to this item: https://hdl.handle.net/2445/171926
Regressió Logísitca Penalitzada
Journal Title
Authors
Director/Tutor
Journal ISSN
Volume Title
Related resource
Abstract
[cat] Actualment, un dels temes d’interès en el món de l’estadística és el Big Data. A l’hora d’estimar un model estadístic amb moltes variables poden sorgir alguns problemes com la multicol·linealitat i la manca d’eficiència, entre d’altres. Tot això, comporta que hi hagi dubtes sobre les estimacions dels paràmetres pels mètodes tradicionals i per aquest motiu s’utilitzen els mètodes de penalització, en concret, en aquest estudi se’n parlarà de tres: Ridge Regression, Lasso Regression i Elastic Net Regression.
Aquests mètodes van ser creats a partir de l’estimació per mínims quadrats dels Models Lineals, però el tipus de model que s’analitzarà en aquest treball és la regressió logística, la qual forma part dels Models Lineals Generalitzats (MLG). Com els MLG s’estimen mitjançant l’estimador de màxima versemblança, aquests mètodes esmentats s’aplicaran d’una manera generalitzada.
[eng] Nowadays, the Big Data is one of a kind issues of interests in the statistics world. Problems as multicollinearity and lack of efficiency, among others, can appear when we pretend to estimate a statistical model with a huge number of variables. By the traditional methods this kind of problems cause doubts about the parameter’s estimation, this is why we use the penalized methods. In this analysis we are going to focus on three of them: Ridge Regression, Lasso Regression and Elastic Net Regression. These methods were created based on the least square estimation of the Lineal Models, but the type of model that we are going to analyze in this project is the logistic regression, which is part of the Generalized Linear Models (GLM). These GLM are estimated by the maximum likelihood estimation, so we will applicate these methods in a general way.
[eng] Nowadays, the Big Data is one of a kind issues of interests in the statistics world. Problems as multicollinearity and lack of efficiency, among others, can appear when we pretend to estimate a statistical model with a huge number of variables. By the traditional methods this kind of problems cause doubts about the parameter’s estimation, this is why we use the penalized methods. In this analysis we are going to focus on three of them: Ridge Regression, Lasso Regression and Elastic Net Regression. These methods were created based on the least square estimation of the Lineal Models, but the type of model that we are going to analyze in this project is the logistic regression, which is part of the Generalized Linear Models (GLM). These GLM are estimated by the maximum likelihood estimation, so we will applicate these methods in a general way.
Description
Treballs Finals de Grau en Estadística UB-UPC, Facultat d'Economia i Empresa (UB) i Facultat de Matemàtiques i Estadística (UPC), Curs: 2019-2020. Tutor: Francesc Carmona Pontaque
Subject (English)
Citation
Citation
MUÑOZ ARAGÓN, Montserrat. Regressió Logísitca Penalitzada. [consulted: 9 of June of 2026]. Available at: https://hdl.handle.net/2445/171926