Python PANDAS: Reindex DataFrameGroupBy Object to Dataframe Error

Question

I have a simple initial dataframe:

ID, ATTRIBUTE
1, thing2
1, thing3
1, thing3
2, thing7
2, thing7
2, thing2
3, thing1
3, thing2

I have a simple groupby object I want to create where I get the mode of ATTRIBUTE (if it is multimodal I call the result 'multithing'):

mode = lambda x: x.mode() if len(x) > 2 else 'multithing'
df_grouped = df.groupby(['ID'], as_index=False)['ATTRIBUTE].agg(mode)

I am trying to get a result after reindexing that looks like this:

ID, ATTRIBUTE
1, thing3
2, thing7
3, multithing

So I can use it like a regular dataframe again and do things like this:

df_final.groupby('ATTRIBUTE')['ID'].count()

MaxU - stand with Ukraine · Accepted Answer · 2017-10-18 22:02:54Z

2

IIUC:

In [203]: df.groupby('ID')['ATTRIBUTE'] \
            .agg(lambda x: x.mode()[0] if len(x.mode()) == 1 else 'multithing')
Out[203]:
ID
1        thing3
2        thing7
3    multithing
Name: ATTRIBUTE, dtype: object

or

In [205]: df.groupby('ID', as_index=False)['ATTRIBUTE'] \
     ...:   .agg(lambda x: x.mode()[0] if len(x.mode()) == 1 else 'multithing')
Out[205]:
   ID   ATTRIBUTE
0   1      thing3
1   2      thing7
2   3  multithing

answered Oct 18, 2017 at 22:02

MaxU - stand with Ukraine

212k37 gold badges403 silver badges437 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

Pylander Over a year ago

Second is perfect. I see what I had been doing wrong with the lambda function previously.

Pylander Over a year ago

One last question if you don't mind? Does the mode lambda in this case ignore np.nan values for ATTRIBUTE? I would want it to only find the mode of the non-null values.

Collectives™ on Stack Overflow

Python PANDAS: Reindex DataFrameGroupBy Object to Dataframe Error

1 Answer 1

2 Comments

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Related