Mixtura finita de una distribución Birnbaum-Saunders basado en la familia de mixtura en parámetros de escala de distribuciones normal asimétrica
Date
2021-10-06
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Pontificia Universidad Católica del Perú
Abstract
La presente tesis muestra la distribución mixtura de distribuciones Birnbaum-Saunders basados en mixturas de escala normal asimétrica (MF-BS-MENA). Este modelo es una extensión a la propuesta de Maehara (2018a) para datos unimodales basados en distribuciones con mixtura de escala normal asimétrica utilizada para modelar datos con percentiles extremos y altamente concentrados a la izquierda de la distribución. El modelo propuesto permite modelar datos con dos o más componentes de mixtura de distribuciones asimétricas como la t de Student asimétrica (TA), la Slash asimétrica (SLA), y la normal contaminada asimétrica (NCA). Para estimar los parámetros del modelo propuesto se presenta un método de estimación basado en el algoritmo de maximización condicional de la esperanza (una extensión del algoritmo EM). Además, se desarrollan simulaciones que muestran la precisión de las estimaciones y los errores estándar. Por último, se realizan aplicaciones con un conjunto de datos reales.
The following thesis presents the nite mixtures of Birbaums-Saunders distributions based on the scale mixture of skew-normal distributions, which are called FM-BS-SMSN. This model is an extension of Maehara (2018a) study of unimodal data based on scale mixture of skew-normal distributions which are used to model extreme percentiles in the left-tail of the distribution. The proposed model can t two or more mixture of components of skewed distributions like Skew-t, Skew-slash, and skew-contaminated normal. The proposed method to accomplish the estimation of model parameters is the expectation of conditional maximization (an extension of EM algorithm). Moreover, simulations and applications are presented to illustrate the robustness of the proposed estimation method and standar errors. Finally, the last chapter presents an aplication for real data sets.
The following thesis presents the nite mixtures of Birbaums-Saunders distributions based on the scale mixture of skew-normal distributions, which are called FM-BS-SMSN. This model is an extension of Maehara (2018a) study of unimodal data based on scale mixture of skew-normal distributions which are used to model extreme percentiles in the left-tail of the distribution. The proposed model can t two or more mixture of components of skewed distributions like Skew-t, Skew-slash, and skew-contaminated normal. The proposed method to accomplish the estimation of model parameters is the expectation of conditional maximization (an extension of EM algorithm). Moreover, simulations and applications are presented to illustrate the robustness of the proposed estimation method and standar errors. Finally, the last chapter presents an aplication for real data sets.
Description
Keywords
Algoritmos, Estadística bayesiana, Estimación de parámetros
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/openAccess