Abstract:

In cases where the dependent variable is count data, it is more appropriate to use Count Data Regression Models in order to produce reliable estimates. These regression models have been developed using discrete distributions since the count data has a discrete structure. The study, it is aimed to model the unemployment duration (in months) of individuals which is count data by using the data of the Survey of Income and Living Conditions (SILC) of the Turkish Statistical Institute (TURKSTAT) in 2019. The independent variables to be used in the analysis are marital status, educational status, general health, and chronic disease have been determined by all possible subset method. The Poisson Regression (PR), Negative Binomial Regression (NBR), Zero Inflated Negative Binomial Regression (ZINB), and Generalized Poisson Regression (GPR) models from count data regression models have been estimated and it has been determined the model best fits data by the information criteria. It has been determined that the model that best fits the data set among the predicted models is the ZINB model.

Keywords: