Estimation of a regression parameters related to exponential survival data of type II censoring

number: 
470
إنجليزية
Degree: 
Imprint: 
Mathematics and Computer Applications
Author: 
Jihan Saleh Ahmed Al-Aggedy
Supervisor: 
Dr Akram M. Al-Abood
year: 
2000
Abstract:

In this thesis , we consider three methods of estimation for a regression coefficients namely maximum likelihood ( ML), weighted least squares ( WLS ), and modified weighted least squares ( MWLS ) , in which the underlay distribution is exponential with censored type II data, where the effect of the explanatory variables on the mean is given by the model μ =exp(χ' β ). Approximation of digamma and digamma functions are used in estimators for MWLS estimators when tables of these functions are not available . Moment properties of the estimators and bias approximation for the ML estimators and the MWLS estimators are given together with the results of a Monte Carlo investigation comparing for the case of a single explanatory variable .