QST:TypeError: sum() got an unexpected keyword argument 'skipna'

[zanwar369]

Got the following error when trying to run a groupby sum after updating to the latest pandas:
TypeError: sum() got an unexpected keyword argument 'skipna'

->for row in Holdings[Holdings.Date>HP].groupby(['a','b','c']).sum(skipna=False)[['Balance','Withdrawal']:
....

[karuppiah7890]

I just checked the help like this

>>> import pandas
>>> pandas.__version__
'1.1.0'
>>> help(pandas.core.groupby.generic.DataFrameGroupBy.sum)

and got

Help on function sum in module pandas.core.groupby.groupby:

sum(self, numeric_only: bool = True, min_count: int = 0)
    Compute sum of group values.

    Parameters
    ----------
    numeric_only : bool, default True
        Include only float, int, boolean columns. If None, will attempt to use
        everything, then use only numeric data.
    min_count : int, default 0
        The required number of valid values to perform the operation. If fewer
        than ``min_count`` non-NA values are present the result will be NA.

    Returns
    -------
    Series or DataFrame
        Computed sum of values within each group.

Looks like skipna is indeed an invalid / unexpected keyword argument. So, seems like a valid error. I also noticed that skipna didn't exist before too, in the argument list, when I checked version 1.0.0 docs and even the oldest docs from 0.20.3

Web docs reference - https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.core.groupby.GroupBy.sum.html

[karuppiah7890]

@zanwar369 ^

[zanwar369]

I just checked the help like this

>>> import pandas
>>> pandas.__version__
'1.1.0'
>>> help(pandas.core.groupby.generic.DataFrameGroupBy.sum)

and got

Help on function sum in module pandas.core.groupby.groupby:

sum(self, numeric_only: bool = True, min_count: int = 0)
    Compute sum of group values.

    Parameters
    ----------
    numeric_only : bool, default True
        Include only float, int, boolean columns. If None, will attempt to use
        everything, then use only numeric data.
    min_count : int, default 0
        The required number of valid values to perform the operation. If fewer
        than ``min_count`` non-NA values are present the result will be NA.

    Returns
    -------
    Series or DataFrame
        Computed sum of values within each group.

Looks like skipna is indeed an invalid / unexpected keyword argument. So, seems like a valid error. I also noticed that skipna didn't exist before too, in the argument list, when I checked version 1.0.0 docs and even the oldest docs from 0.20.3

Web docs reference - https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.core.groupby.GroupBy.sum.html

Hmm interesting--thank you! Does anyone know why this was removed? It was functioning in previous versions perfectly

[karuppiah7890]

That's weird. I don't know, it feels like the previous versions didn't have this. Could you tell which version you used that had this? So that it can be confirmed? Maybe the old docs just didn't reflect what was in the code

[zanwar369]

Sorry for the delay-- it works perfectly on 0.25.3
Looks like it stopped in version 1.0.0. i was utilizing this since I noticed a distinct difference in sums, for very large calculations with and without this qualifier.

[karuppiah7890]

Ah okay

[simonjayhawkins]

Thanks @zanwar369 for the question. I think is is covered by #29481, see #29481 (comment) for a workaround

[zanwar369]

Thanks @zanwar369 for the question. I think is is covered by #29481, see #29481 (comment) for a workaround

Is there an actual fix in the works? Confused as to why it was removed in the first place

[jorisvandenbossche]

The underlying issue here is that the skipna keyword is at the moment not yet implemented for groupby reductions like groupby(..).sum().

It might be that before this keyword was ignored and recently started to raise, but it never actually worked (or was never documented).

The improvement to add skipna to the grouped reductions is covered in #15675, so going to close this issues.

[jorisvandenbossche]

Duplicate of #15675