Estimación de la probabilidad de anemia infantil usando un modelo de regresión skew-probit
Date
2024-11-20
Journal Title
Journal ISSN
Volume Title
Publisher
Pontificia Universidad Católica del Perú
Abstract
La desnutrición crónica infantil en menores de cinco años de edad es un problema de salud pública en el Perú. Modelos estadísticos apropiados pueden ayudar a identificar variables o factores que permitan estimar la probabilidad de anemia infantil. Los modelos estadísticos para este tipo de datos binarios más conocidos son el modelo de regresión logística y probit. En esta tesis se aplican estos modelos y el modelo skew-probit, una extensión del modelo probit cuya función de enlace es asimétrica, en particular usando una versión estandarizada de la distribución skew normal. La inferencia se realiza a través del enfoque bayesiano, específicamente a través de la aproximación de Laplace integrada y anidada (INLA) debido a su eficiencia computacional. Cabe resaltar que se usa una distribución a priori penalizada compleja (PC prior) para el parámetro de sesgo de la skew normal, de esta forma se “cuantifica” la elección del modelo skew-probit respecto al modelo probit. Los resultados obtenidos para la estimar la probabilidad de anemia en niños menores de cinco años justifican la elección del modelo skew-probit.
Chronic childhood malnutrition in children under five years of age is a public health problem in Peru. Appropriate statistical models can help to identify variables or factors that allow estimating the probability of childhood anemia. The best-known statistical models for this type of binary data are the logistic regression and probit models. In this thesis, these models and the skew-probit model, an extension of the probit model whose link function is asymmetric, are applied, in particular using a standardized version of the skew normal distribution. Inference is performed through the Bayesian approach, specifically through integrated nested Laplace approximation (INLA) due to its computational efficiency. It should be noted that a complex penalized prior distribution is also used for the skew parameter, through this approach the choice of the skew-probit model is “quantified” with respect to the probit model. The results obtained to estimate the probability of anemia in children under five years of age justify the choice of the skew-probit model.
Chronic childhood malnutrition in children under five years of age is a public health problem in Peru. Appropriate statistical models can help to identify variables or factors that allow estimating the probability of childhood anemia. The best-known statistical models for this type of binary data are the logistic regression and probit models. In this thesis, these models and the skew-probit model, an extension of the probit model whose link function is asymmetric, are applied, in particular using a standardized version of the skew normal distribution. Inference is performed through the Bayesian approach, specifically through integrated nested Laplace approximation (INLA) due to its computational efficiency. It should be noted that a complex penalized prior distribution is also used for the skew parameter, through this approach the choice of the skew-probit model is “quantified” with respect to the probit model. The results obtained to estimate the probability of anemia in children under five years of age justify the choice of the skew-probit model.
Description
Keywords
Anemia en niños, Análisis de regresión logística, Probabilidades
Citation
Collections
Endorsement
Review
Supplemented By
Referenced By
Creative Commons license
Except where otherwised noted, this item's license is described as info:eu-repo/semantics/embargoedAccess