Share
مقارنة طرائق تقدير معلمات الانـموذج الثنائي اللوجستك المختلط باستخدام المحاكاة مع تطبيق عملي == Comparing The Estimation Methods of The Parameters of The Logistic Linear Mixed Model By Using The Simulation With Practical Application
Author name:
هند وليد عبد الرحمن الجبوري
Supervisor name:
ظافر حسين رشيد النجار
General topic:
Administration and Economics
Specific topic:
Statistics
Degree:
Master
University:
University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language:
Arabic
University location:
Baghdad
First pages:
07T4153 - p.pdf
Abstract:
تم في هذا البحث دراسة احد اهم النماذج الواسعة الاستخدام والتطبيق في تحليل البيانات التي تاخذ شكل تجمعات Clustered والتي تكون ذات استجابات مرتبطة Correlated وهو الانموذج الخطي العام المختلطGeneralized Linear Mixed Model (GLMM) لتحليل البيانات الطولية Long | This research was conserning in the study of one of the important models that are widely used in analyzing the data which take clustered form and have correlated responses, this model is Generalized Linear Mixed Model (GLMM) for analyzing the longitudinal data which it's responses are correlated. So this research was dealing with this model in an expanded form including its importance, uses, feature, modeling and the estimation methods then focusing on one of the most widely used examples when the responses of longitudinal Data are Binary which is Binary Logistic Mixed Model taking into account its modeling, importance, uses and the parameters estimation methods, so three important estimation methods were used to estimate the fixed and random effects parameters and these are : 1) Classical maximum likelihood (ML) Method.That include three basic algorithm by which the maximum likelihood estimator is obtained and it's as follows : I) Monte Carlo Newton Raphson (MCNR).II) Monte Carlo Expectation Maximization (MCEM).III) Simulation Maximum Likelihood (SML).2) Robust Maximum Likelihood (RML) Method.3) Penalized - Quasi Likelihood (PQL) Method.In the experimental aspect comparison was done of which is the best among these methods through the simulation procedure by using Monte Carlo method and implementing several experiments using two of the important statistical measures which are Mean Square Error (MSE) and Bias, generally as a result it was found that Robust Maximum Likelihood (RML) Method is the best between these methods as it has minimum Mean square error and minimum bias comparing with the other methods.While in the application aspect practical application was done on data represent the successive monthly measurements for diebetic children whom depend on the insulin treatment which represent the fixed effect and the patient represents the random effect in order to study the effect of both the insulin dose and the patient on the blood sugar rate, it was found that the insulin dose has significant effect on the blood sugar rate while the patient has not that effect
