Pandas, group by with max return AssertionError:

Carlos Carvalho :

There's something wrong with pandas, and I would like your opinion,

I've this Dataframe where I need to get the max values, code is just below,

df_stack=pd.DataFrame([[1.0, 2016.0, 'NonResidential', 'Hotel', 98101.0, 'DOWNTOWN',
        47.6122, -122.33799, 1927.0, 57.85220900338872,
        59.91269863912585],
       [1.0, 2016.0, 'NonResidential', 'Hotel', 98101.0, 'DOWNTOWN',
        47.61317, -122.33393, 1996.0, 55.82342114189166,
        56.86951201265458],
       [3.0, 2016.0, 'NonResidential', 'Hotel', 98101.0, 'DOWNTOWN',
        47.61393, -122.3381, 1969.0, 76.68191235628086,
        77.37931271575705],
       [5.0, 2016.0, 'NonResidential', 'Hotel', 98101.0, 'DOWNTOWN',
        47.61412, -122.33664, 1926.0, 68.53505428597694,
        71.00764283155655],
       [8.0, 2016.0, 'NonResidential', 'Hotel', 98121.0, 'DOWNTOWN',
        47.61375, -122.34047, 1980.0, 67.01346098859122,
        68.34485815906346]], columns=['OSEBuildingID', 'DataYear', 'BuildingType', 'PrimaryPropertyType', 
 'ZipCode', 'Neighborhood', 'Latitude', 'Longitude', 'YearBuilt', 
 'SourceEUI(KWm2)', 'SourceEUIWN(KWm2)' ])

When I run the code below :

df_stack[['OSEBuildingID', 
          'DataYear', 
          'BuildingType', 
          'PrimaryPropertyType', 
          'ZipCode', 'Neighborhood', 'Latitude', 'Longitude', 
          'YearBuilt', 'SourceEUI(KWm2)', 'SourceEUIWN(KWm2)']].groupby('OSEBuildingID').max()

I get an error, "AssertionError: " the same you'll probably get if you try this. But, when I comment this two columns and I run the code again

df_stack[['OSEBuildingID', 
          'DataYear', 
          #'BuildingType', 
          #'PrimaryPropertyType', 
          'ZipCode', 'Neighborhood', 'Latitude', 'Longitude', 
          'YearBuilt', 'SourceEUI(KWm2)', 'SourceEUIWN(KWm2)']].groupby('OSEBuildingID').max()

I get the results

     DataYear  ZipCode Neighborhood  Latitude  Longitude  YearBuilt  SourceEUI(KWm2)  SourceEUIWN(KWm2)
OSEBuildingID                                                                                                    
1.0              2016.0  98101.0     DOWNTOWN  47.61317 -122.33393     1996.0        57.852209          59.912699
3.0              2016.0  98101.0     DOWNTOWN  47.61393 -122.33810     1969.0        76.681912          77.379313
5.0              2016.0  98101.0     DOWNTOWN  47.61412 -122.33664     1926.0        68.535054          71.007643
8.0              2016.0  98121.0     DOWNTOWN  47.61375 -122.34047     1980.0        67.013461          68.344858

If I replace max() by mean() I can uncomment those 2 lines and runc the code with no problem. This behaviour it only happens with max() and min(), well I just test max, mean and min, But I need to get the max.

Thank you if can help.

ALollz :

This was a regression in 1.0.0 that was fixed with '1.0.1', so I suggest you upgrade your version.

Fixed regression in .groupby().agg() raising an AssertionError for some reductions like min on object-dtype columns (GH31522)