Maybe abandon SaQC.roll and others/Bug in SaQC.roll
I encountered several bugs when applying saqc.roll
on not-harmonized data.
This is because of some conditioning branching reasons when trying to set center
to False and also because of a bug when center
is True - it has to do with some data constellation that i will provide in a followup issue...
What i would like to push forward here: i think we should/could just abandon our own rolling implementations in favor of something like (wich already works fine):
qc = qc.processGeneric(field, func=lambda x:x.rolling(**params))
i have the impression the generic wrappers are central to any slightly more complex qc pipeline any ways, since the recently implemented source target mechanisms actually enables them of doing everything a real processing function might be able of ... but with a custom numerical function ... and also, there is no reason to have time series data manipulation algorithms concurring with current pandas implementations...