پديد آورنده :
يگانه، مهسا
عنوان :
رويكرد نيم پارامتري بيزي به انتخاب اثرات تصادفي در مدل هاي با اثرات آميخته
مقطع تحصيلي :
كارشناسي ارشد
گرايش تحصيلي :
آمار اقتصادي و اجتماعي
محل تحصيل :
اصفهان: دانشگاه صنعتي اصفهان، دانشكده علوم رياضي
صفحه شمار :
نه، [۷۰]ص.: مصور، جدول، نمودار
استاد راهنما :
ريحانه ريخته گران
واژه نامه :
انگليسي به فارسي; فارسي به انگليسي
توصيفگر ها :
داده هاي طولي , مدل هاي با اثرات آميخته , انتخاب اثرات تصادفي , رويكرد بيز , فرآيند ديريكله
استاد داور :
هوشنگ طالبي، ايرج كاظمي
تاريخ ورود اطلاعات :
1396/12/13
چكيده انگليسي :
A Bayesian semi parametric approach to random e ects selection in mixed e ects models Mahsa Yeganeh mahsa yeganeh@math iut ac ir 2017 Department of Mathematical Sciences Isfahan University of Technology Isfahan 84156 83111 Iran Supervisor Dr Reyhaneh Rikhtehgaran bomoomi@cc iut ac ir Advisor Dr Zahra Saberi romidi@cc iut ac ir 2010 MSC 62J12 62G 08 Keywords Longitudinal data Mixed e ects models Random e ects Bayesian approach Dirichletprocess AbstractNowadays with developing technology it is possible to collect and record a large number of variablesfor the underlying experimental units in di erent researches If these variables are collected overtime for di erent individuals the resulting data is called longitudinal data Longitudinal studies areoften used in Econometric Medicine and social sciences One important source of dependency inthese types of data sets is the intra class correlation which is created due to the e ects of individuals characteristics on their corresponding observations Mixed e ects models are regression models which are considered to take into account the intra classcorrelation in longitudinal data through considering random e ects in the structure of the underlyingmodels In mixed e ects models the problem of random e ects selection is one of the importantissues speci cally when the number of explanatory variables is large For this purpose in this thesis at rst we describe a general classi cation of methods for random e ects selection in linear mixed e ects models These methods are generally divided into four cate gories subset selection methods shrinkage methods hypothesis testing methods and Bayesian meth ods of variable selection Also the advantages and disadvantages of these methods are discussed Among these selection methods in this thesis we use the Bayesian hierarchical method for the random e ects selection In this method at rst the Cholesky decomposition of the variance covariance matrix
استاد راهنما :
ريحانه ريخته گران
استاد داور :
هوشنگ طالبي، ايرج كاظمي
