You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Unstacking changes the data type to float for NaN's.
Are there thoughts on alternative options, e.g. fill_value=0 or return_boolean_mask, in order to retain the original data type?
Currently, I obtain a boolean missing array by checking for isnan.
Then I call fillnan(0) and convert the data type back to integer.
However, this is quite inefficient.
Activity
shoyer commentedon Nov 13, 2019
We should definitely have a
fill_valueoption here, and ideally asparseoption, too.Conceptually
unstackis very similar tofrom_dataframe.