2893

2723

كارشناسي ارشد

آمار كاربردي

اصفهان: دانشگاه صنعتي اصفهان، دانشكده علوم رياضي

1384

ده، 139، [I]ص.: مصور، جدول، نمودار

1396/10/26

كتابنامه

علوم رياضي

رياضي

ID2723

به فارسي و انگليسي: قابل رويت در نسخه ديجيتالي

Abstract There are many industrial experiments where the responsevariable is nonnormal Tradition ally variance stabilizing transformation are made on such a response in order to obtain conditionsneeded to use ordinary least squares Generalized linear models GLMs offer a powerful alternativeto data transformation Specifically the performance in response estimation and prediction for aGLM is often superior to the model built using data transformation In Generalized linear models confidenceintervals are based on asymptotic theory As such they are regarded as statistically validonly for large samples Therefore in order to use confidence intervals to compare models theseasymptotic intervals evalute in terms of coverage The results show that the coverageobserved inthe Monte Carlo simulation for small samples is very close to the asymptotic theoritical claim of95 therfore confidence intervals lengths are an oppropriate way to compare the GLM approchwith other approches Finally with several additional real examples of designed experiments withnonnormal response give an explicit comparison of the potential benefits of the GLM relative to theconventional approch based on data transformation d 1 j I U J V U JIJ Aq

علي زينل همداني

سروش علمرادي

هوشنگ طالبي، سعيد پولاد