Hm, can you refine your question? In how far does Apply Model improve the classification?
Ok, in you case you are applying the model on the training data. You could use a X-Prediction - it works the same way as the cross validation, but it delivers the predicted values instead of the performance.
Thanks for your advice. But after testing the operator as you can see in the picture, now my question is why are the confusion matrices different, although I used the same operators in X-Validation and X-Prediction?