Publication: Genelleştirilmiş Lineer Karma Modellerde Bayesci Yaklaşımın Kullanımı
Loading...
Date
Authors
Journal Title
Journal ISSN
Volume Title
Abstract
Günümüzde çok yaygın kullanım alanlarına sahip olan genelleştirilmiş lineer karma modeller (GLKM), bu çalışmanın temelini oluşturmaktadır. GLKM, karma modellerin özel bir halidir. GLKM lineer tahmin edicilerin genel sabit etkilere rasgele etkilerin eklenmesiyle genelleştirilmiş lineer modellerin (GLM) bir genişlemesidir. Rasgele etkilerin genellikle normal dağılmlı olduğu varsayılır. Zaten, 'Genelleştirilmiş' kelimesi ile bağımlı değişkenin sadece normal dağılıma sahip olmadığı, 'Karma' kelimesiyle de modeldeki genel sabit etkilere rasgele etkilerin eklenmesi ifade edilmektedir. Parametre tahmininde ve model seçiminde uzman görüşlerin modele katılmasını öngören Bayesci yaklaşımların kullanımı da son yıllarda oldukça popüler hale gelmiştir. Bu çalışma, GLKM ile Bayesci yaklaşımın birleştirilmesini ve GLKM' de parametre tahmin yöntemleri için kullanılan klasik istatistiksel yöntemler ve Bayesci yöntemlerden bahsedilmiştir. Bayesci yaklaşımı klasik yaklaşımdan ayıran en önemli nokta, parametreler hakkında önsel bilgi kullanımı olduğundan bu çalışmada farklı önseller kullanılmış ve bu önseller hem parametre tahmini hem de model uyumu anlamında karşılaştırılmıştır. Klasik ve Bayesci GLKM yöntemleri, farklı doktorların akciğer kanseri hastalarının aldıkları tedaviden sonraki iyileşme durumlarını etkileyen faktörleri ölçmek için yapılan bir çalışmaya ait veriye uygulanmıştır. Veride farklı hasta sayısına sahip 407 doktor arasından 40 hastası bulunan 9 doktor rasgele seçilmiştir. Doktorlar rasgele etkili parametre ve hastalardan alınan 6 sabit ölçüm değeri ise sabit etkili parametre olarak modele katılmıştır.
Nowadays, having a very widespred use generalized linear mixed models (GLMM) is the basis of this study. A generalized linear mixed model is a particular type of mixed model. It is an extension to the generalized linear model in which the linear predictor contains random effects in addition to the usual fixed effects. These random effects are usually assumed to have a normal distribution. 'Generalized' with the words that were not normally distributed dependent variable only, 'mixed' with the word general fixed effects and random effects in the model is the inclusion of expression. In the parameter estimation and model selection, the use of Bayesian approach that envisage to participate in the expert opinion has become quite popular in recent years. In this study, generalized linear mixed model with Bayesian approach for combining and is intended to be introduced classical statistical methods and Bayesian methods used for parameter estimation methods. Different priors have used and this priors are compared in terms of parameter estimation and model adaptation because Bayesian approach that distinguishes it from the classical approaches, the most important point is the use of a priori information about the parameters. In this study has used a study data that wants to know what patient and physician factors are most related to whether a patient's lung cancer goes into remission after treatment as part of a larger study of treatment outcomes and quality of life in patients with lunger cancer. Classical and Bayesian GLMM methods have been applied. It was randomly selected 40 patients with nine doctors from 407 doctors with different numbers of patients. doctors are random effects parameter and 6 fix measured value from patients are fixed effect parameters.
Nowadays, having a very widespred use generalized linear mixed models (GLMM) is the basis of this study. A generalized linear mixed model is a particular type of mixed model. It is an extension to the generalized linear model in which the linear predictor contains random effects in addition to the usual fixed effects. These random effects are usually assumed to have a normal distribution. 'Generalized' with the words that were not normally distributed dependent variable only, 'mixed' with the word general fixed effects and random effects in the model is the inclusion of expression. In the parameter estimation and model selection, the use of Bayesian approach that envisage to participate in the expert opinion has become quite popular in recent years. In this study, generalized linear mixed model with Bayesian approach for combining and is intended to be introduced classical statistical methods and Bayesian methods used for parameter estimation methods. Different priors have used and this priors are compared in terms of parameter estimation and model adaptation because Bayesian approach that distinguishes it from the classical approaches, the most important point is the use of a priori information about the parameters. In this study has used a study data that wants to know what patient and physician factors are most related to whether a patient's lung cancer goes into remission after treatment as part of a larger study of treatment outcomes and quality of life in patients with lunger cancer. Classical and Bayesian GLMM methods have been applied. It was randomly selected 40 patients with nine doctors from 407 doctors with different numbers of patients. doctors are random effects parameter and 6 fix measured value from patients are fixed effect parameters.
Description
Tez (doktora) -- Ondokuz Mayıs Üniversitesi, 2014
Libra Kayıt No: 111056
Libra Kayıt No: 111056
Keywords
Citation
WoS Q
Scopus Q
Source
Volume
Issue
Start Page
End Page
117
