عنوان :
طرح هاي بلوكي D- بهينه براي پاسخ هاي همبسته غير نرمال
مقطع تحصيلي :
كارشناسي ارشد
محل تحصيل :
اصفهان: دانشگاه صنعتي اصفهان، دانشكده علوم رياضي
صفحه شمار :
يازده،93ص.: جدول،نمودار(رنگي)
يادداشت :
ص.ع.به فارسي و انگليسي
استاد راهنما :
سعيد پولادساز، ايرج كاظمي
توصيفگر ها :
متغير پاسخ دودويي , طرح بلوكي دقيق , D- بهينگي بيزي , مدل هاي خطي تعميم يافته , طرح استوار
تاريخ نمايه سازي :
10/4/93
استاد داور :
هوشنگ طالبي، ريحانه ريخته گران
چكيده انگليسي :
D optimality Block Designs for Experiments With Correlated Non Normal Response Najmeh Ghasri n ghasri@math iut ac ir 2014 Department of Mathematical Sciences Isfahan University of Technology Isfahan 84156 83111 Iran First Supervisor Dr Saeed Pooladsaz spooladsaz@cc iut ac ir Second Supervisor Dr Iraj Kazemi kazemi@cc iut ac ir 2010 MSC 62J 99 62K05 62K10 62K99 Keywords Binary response Exact block designs D optimality Generalized Linear Model Gen eralized Estimating Equations Robust design AbstractMany experiments in science and engineering aim to understand a process by modeling a responsevariable using a generalized linear model GLM with several explanatory variables By researchersMethods developed for constructing e cient and optimal designs for such experiments using homo geneous experimental units Also they demonstrated the potentially poor performance of standardlinear model designs for GLMs In this thesis we develop methods for constructing designs wherethe heterogeneous units are grouped into homogeneous sets or blocks to increase the accuracy of in ferences made from the experimental data We focus on the exponential family of distributions andmodels for the marginal response which in the data analysis are tted by the method of generalizedestimating equations GEE The GEE models provide a computationally e cient extension of GLMsby allowing dependence between observations from units in the same block characterized throughan intra block working correlation structure and have been widely applied in the social and medicalsciences Similar to GLMs and other nonlinear models the design of an experiment for GEE models iscomplicated by the dependence of design performance on the values of the model parameters Locallyoptimal designs for given values of the model parameters are e ective when strong prior informationis available from pilot experiments or historical data or where practicable as part of a sequential
استاد راهنما :
سعيد پولادساز، ايرج كاظمي
استاد داور :
هوشنگ طالبي، ريحانه ريخته گران
