Variable selection methods in cox regression model

In survival analysis, the Cox regression model is used to model the relationship between survival time and one or more covariates. When modeling, it is important to determine which covariates should or should not be included in the linear component of the Cox regression model. Stepwise selection methods, best subset selection methods and shrinkage methods are used to determine these variables. While stepwise selection methods and best subset selection methods are frequently used in the literature, the usage of shrinkage methods has gained importance in recent years. In the thesis, general information about the literature for variable selection methods in the Cox regression model is given; Forward selection, backward elimination, stepwise forward selection, stepwise backward elimination and augmented backward elimination methods for stepwise selection methods; Akaike information criterion and Bayesian information criterion for best subset methods; Ridge regression model and LASSO regression model were examined in detail for shrinkage methods. These methods were applied on the kidney cancer dataset in the literature, and the results were interpreted. Keywords: Cox regression models, Variable selection, Stepwise selection, Shrinkage methods