Calculated channels uses masked data

[mayankchetan]

When creating new calculated channels after masking the data (eg: {B1TipTDxr} - np.mean({B1TipTDxr}) ), the value of np.mean({B1TipTDxr}) is calculated from the whole time history.

Eg: This can affect removing means from simulations which have long transients.

[ebranlard]

Thanks, I can see how that can be a problem. The formula is indeed evaluated on the whole dataframe (df) before the mask is applied:

pyDatView/pydatview/formulae.py

Line 16 in 4140c24

sFormula=sFormula.replace('{'+c_no_unit+'}','df[\''+c_in_df+'\']')

The mask is applied when the "columns" are retrieved:

pyDatView/pydatview/Tables.py

Line 874 in 4140c24

c = self.data.iloc[self.mask, i]

There is likely a better way to apply a mask to a pandas dataframe without duplicating the memory, but it might require some changes.

Currently, I don't see an easy path forward, but some workaround:

• You can "+ Add" after the mask, and then add columns on the newly added tables...
• You can experiment with the new features added to "Min-Max" plot, that might actually work the way you intend to. Let me know if the "min-max" feature (e.g. with Y-centering = "mean=0") doesn't remove the "masked-mean".