بعض طرائق التقدير لمعلمة القياس لتوزيع لابلاس : دراسة مقارنة == Some Baye,s Estimators for the Scale Parameter of the Laplace Distribution / A Comparison Study
Author name:
عماد فرهود محي الشريفي
Supervisor name:
هدى عبد الله رشيد
General topic:
Mathematics
Specific topic:
Mathematics
Degree:
Master
University:
Mustansiriyah University - College Of Science - Department Of Mathematics
Language:
English
University location:
Baghdad
First pages:
27T1170 - p.pdf
Abstract:
The Laplace distribution is important to the development of mathematics ,statistics, physics, and astronomy, and usually used as its characteristics apply to a large ratio of natural, social and economic phenomenon. Hence, the importance of this thesis is to Find the best estimator for the scale parameter θ, when location parameter (a) is known and unknown through obtaining and comparing the classical estimators (Maximum Likelihood Estimator, Uniformly Minimum variance unbiased Estimator, and Minimum Mean Squared Error), as well as Bayesian estimators under different loss functions (Quadratic loss function, squared - Log error loss function, Generalized squared error loss function, Entropy loss function, Suggested loss function).In order to get a better understanding of our Bayesian analysis, we consider the non - informative prior for the scale parameter () using Jeffreys prior information, as well as informative prior density represented by Gumbel Type II Prior.All these estimators are compared empirically using Mont - Carlo simulation by employing the mean squared errors (MSE's).Also we derived the Minimax estimators of the scale parameter θ for the Laplace distribution for all loss functions referred to above and reach to the estimators under Quadratic loss function, Squared - log error loss function, special case of Suggested loss function and Entropy loss function are Minimax estimators.Among conclusions that have been reached, the Suggested loss function with Gumble prior has recorded the best performance of the used for other loss functions. On the other hand, Gumble prior under different loss functions with Gumble Type II prior is better comparing to the corresponding Jeffrey's prior (under the same loss function) for all cases.
