Unbalanced data

[jiaxi98]

As our simulation always starts from unsteady simulation and later transit to steady simulation which may influences the constitutent of the training dataset (unbalanced dataset). Maybe we should only use those data that has already come into the steady state regime?

[jiaxi98]

We could compare the dataset with the whole simulation from unsteady state with the dataset truncated from the period where the fluid is already steady. But this will not be the priority job