Document Details

Document Type : Thesis 
Document Title :
LOG EXPONENTIAL-PARETO REGRESSION MODELS
لوغاريثم نماذج انحدار باريتو الأسية
 
Subject : Faculty of Science 
Document Language : Arabic 
Abstract : A new odds exponential-Pareto IV distribution is defined and investigated.Also, log exponential-Pareto and log odds exponential-Pareto IV distributions are introduced.Some of their properties are derived.Moreover, two new location-scale regression models are constructed;one based on the log exponential-Pareto and the other based on the log odds exponential-Pareto IV distributions.The maximum likelihood method is applied to estimate the parameters of the odds exponential-Pareto IV distribution.Also, the parameters of the proposed regression models are estimated using two different methods for censored data. First, the maximum likelihood estimators are obtained along with their asymptotic variance covariance matrices and confidence intervals.Additionally, jackknife method is considered as another method to estimate the unknown parameters.The sensitivity analysis are applied to detect the possible influential observations.Also, martingale and deviance residuals are explored to assess the adequacy of the proposed model and to detect outliers.The efficiency of maximum likelihood estimators for the parameters of odds exponential-Pareto IV distribution is investigated using some Monte Carlo simulation studies.Also,the performance of the maximum likelihood estimate is compared with the jackknife estimate through some Monte Carlo simulation studies for the proposed regression models.The simulation results showed that the jackknife estimates are more efficient than the maximum likelihood estimates based on average of standard error.The martingale and deviance residuals properties are computed and the empirical distributions are plotted.Finally,the real-data sets are analyzed under the odds exponential-Pareto IV distribution to show the potentiality of this new distribution.Also, the flexibility of the regression models are illustrated by mean of some real-life applications.The results demonstrated the superior performance of the proposed models. 
Supervisor : Dr. Lamya A. Baharith 
Thesis Type : Master Thesis 
Publishing Year : 1441 AH
2020 AD
 
Co-Supervisor : Dr. Hadeel S. Klakattawi 
Added Date : Saturday, June 20, 2020 

Researchers

Researcher Name (Arabic)Researcher Name (English)Researcher TypeDr GradeEmail
خلود معيوض البلاديAL-Beladi, Kholod MayoudResearcherMaster 

Files

File NameTypeDescription
 46441.pdf pdf 

Back To Researches Page