Determination smoking habits by logistic regression analysis: R application

The response variable is assumed normally distributed in Linear and nonlinear regression models. But in practice, this assumption often does not occur. In this case, one of the appropriate model to be used is generalized linear model (GLM). GLM takes into account two assumption: Response variables are independent and has a distribution from the exponential family.If the response variable has a binomial distribution, logistic regression analysis is used. Similarly, if the response variable has a Poisson distribution, Poisson regression analysis is used. The main purpose is to obtain, the simplest model that is best to explain the causation between dependent variable and independent variables of causation is obtained.Generalized linear models can be seen as a combination of linear and nonlinear regression models. GLM, doğrusal ve doğrusal olmayan regresyon modellerinin bir birleşimi olarak görülebilir. Inference models and model adaptation can be performed under the same framework. In fact, this combined approach, widely used and easy to use computer software that is supported by. One of these software is R. The first use of the generalized linear model was limited to health sciences applications rapidly increased in other scientific fields. At this point, our study, focuses on the use of generalized linear models in the social sciences and these applications to be made in R.In this study, basically, the properties of generalized linear models and exponential families are touched upon and logistic and Poisson regression analysis was investigated. In the application part of this thesis, in the Inonu University, Faculty of Economic and Administrative Sciences, the factors affecting the smoking habits of students is revealed. In addition to smoking ban applied in our country of young people to look out and the effect of the ban on smoking to evaluate intended. Logistic regression analysis was used for this purpose and for the R program is used effectively.Anahtar Kelimeler: Generalized Linear Model, Logistic Regression, Poisson Regression, Smoking Habit