No, because your training data is only an appropriation of the actual workload.