This master thesis aims at deriving, discussing and comparing the Vuong Test for model selection as well as three modifications for it. On the one hand, we compare the test strategies theoretically with respect to the assumptions required for their application as well as their asymptotic size. On the other hand, we evaluate the test strategies with respect to their performance regarding various applications. For this purpose, several examples are examined within a simulation study. We will highlight the weaknesses of the classical Vuong test and show that the modifications provide the potential to overcome them. In a further step, we motivate a model selection test for Support Vector Regression models theoretically and examine our proposed approach experimentally in terms of a simulation study. We can show that the experiments support our hypothesis regarding the asymptotic distribution of the test statistic in the discussed examples.      «

This master thesis aims at deriving, discussing and comparing the Vuong Test for model selection as well as three modifications for it. On the one hand, we compare the test strategies theoretically with respect to the assumptions required for their application as well as their asymptotic size. On the other hand, we evaluate the test strategies with respect to their performance regarding various applications. For this purpose, several examples are examined within a simulation study. We will highl...     »