I am doing a churn prediction task with class 1 (churn) is the majority class.
Result for both classes
My question is: Is this result a consequence of class setting (usually positive class is the negative class). And how should I interpret the result above? Is there any trade-off between the class 0 and class 1 in terms of F1 score?

In the result, the default F1 had better score with the imbalanced dataset, while I was expecting the resampling methods would improve the F1. They actually improved F1 for the minority class, but not the majority class.