%0 Journal Article
%T Binomial Additive Modeling for Nonlinear Relationships
%A Ahmet SEZER
%A Bet¨šl KAN
%A Berna YAZICI
%J Turkiye Klinikleri Journal of Biostatistics
%D 2011
%I Turkiye Klinikleri
%X Objective: In most of the fields, researcher misapplies linear models, although there are more complex models needed. In this study generalized additive modeling of a real life data set is taken into account for binomial response case. Material and Methods: Generalized additive models (GAM) have become an elegant and practical option in model building. Those models represent an extension of generalized linear models (GLM) with a linear predictor involving a sum of smooth functions of covariates. GAMs allow for rather flexible specification of the dependence of the response on the covariates by specifying the model in terms of smoothing functions besides linear components. The advantage in that type of modeling is that the forms of the explanatory variables are not predetermined unlike in linear regression modeling but are constructed according to information derived from the data. This provides significant advantage for the nonparametric modeling over the linear regression modeling. Results: The models that are constructed for different components are interpreted and compared using Un-Biased Risk Estimator criterion. The model which best explains the structure of the dataset using splines in term of UBRE is given.
%K Linear models
%K models
%K statistical
%K models
%K theoretical
%U http://www.turkiyeklinikleri.com/pdf/?pdf=23298314f5e56cb4fbe82ea0122fc6b8