But isn't it possible to design multiple models, where the judgement of the "best model" is dependent on your goals (e.g. reduction of false positives vs false

But isn't it possible to design multiple models, where the judgement of the "best model" is dependent on your goals (e.g. reduction of false positives vs false negatives)?