Asymptotic distribution of Crámer-von Mises and Anderson-Darling tests for right-censored data

The aim of this work is to study the asymptotic distribution of the Cramér-von Mises (C-vM) statistic for tests of goodness of-fit, when the data are subject to random right censorship, as well as the implementation of the test in R. We introduce the framework of goodness-of-fit tests and provide an...

Descripción completa

Detalles Bibliográficos
Autor: Tapia Pérez, Clara
Tipo de recurso: tesis de maestría
Fecha de publicación:2022
País:España
Institución:Universitat Politècnica de Catalunya (UPC)
Repositorio:UPCommons. Portal del coneixement obert de la UPC
Idioma:inglés
OAI Identifier:oai:upcommons.upc.edu:2117/371949
Acceso en línea:https://hdl.handle.net/2117/371949
Access Level:acceso abierto
Palabra clave:Survival analysis (Biometry)
Goodness-of-fit
Empirical distribution function
Cramér-von Mises
Right-censored
Survival analysis
Asymptotic distribution
Implementation
Anàlisi de supervivència (Biometria)
Classificació AMS::62 Statistics::62N Survival analysis and censored data
Àrees temàtiques de la UPC::Matemàtiques i estadística::Estadística matemàtica
Descripción
Sumario:The aim of this work is to study the asymptotic distribution of the Cramér-von Mises (C-vM) statistic for tests of goodness of-fit, when the data are subject to random right censorship, as well as the implementation of the test in R. We introduce the framework of goodness-of-fit tests and provide an exhaustive review of the ones based on the empirical distribution function, for both complete and right-censored data. We explain the derivation of the asymptotic distribution of the C-vM statistc under a particular model of random censorship, following the results presented by Koziol and Green in 1976. Finally, we implement the test in R and conduct a simulation study to check its performance.