pandas groupby 将相同的函数应用于多列

Question

I have a dataframe that looks like:

>>> df = pd.DataFrame({'id':[1,1,1,1,1,2,2,2],'month':[1,1,2,2,2,1,2,2],'value1':[1,1,3,3,5,6,7,7], 'value2': [9,10,11,12,12,14,15,15], 'others': range(8)})
>>> df

   id  month  value1  value2  others
0   1      1       1       9       0
1   1      1       1      10       1
2   1      2       3      11       2
3   1      2       3      12       3
4   1      2       5      12       4
5   2      1       6      14       5
6   2      2       7      15       6
7   2      2       7      15       7

I want to do perform a custom function whose input is a series on value1 and value2 :

def get_most_common(srs):
    """
    Returns the most common value in a list. For ties, it returns whatever
    value collections.Counter.most_common(1) gives.
    """
    from collections import Counter

    x = list(srs)
    my_counter = Counter(x)
    most_common_value = my_counter.most_common(1)[0][0]

    return most_common_value

Expected result:

               value1    value2
   id  month   
   1      1       1       9
          2       3      12
   2      1       6      14
          2       7      15

The function was written like that because initially I only had to apply it to a single column ( value1 ) so df = df.groupby(['id,'month'])['value1'].apply(get_most_common) worked. Now I have to apply it to two columns simultaneously.

Attempts:

1. Apply

df = df.groupby(['id,'month'])[['value1','value2']].apply(get_most_common) gave:

id  month
1   1        value1
    2        value1
2   1        value1
    2        value1

2. Transform

df = df.groupby(['id,'month'])[['value1','value2']].transform(get_most_common)

gave this

   value1  value2
0       1       9
1       1       9
2       3      12
3       3      12
4       3      12
5       6      14
6       7      15
7       7      15

3. applymap doesn't work.

What am I missing here?

Answer 1

Use GroupBy.agg - it run function for each column separately:

df = df.groupby(['id','month'])['value1','value2'].agg(get_most_common)
print (df)
          value1  value2
id month                
1  1           1       9
   2           3      12
2  1           6      14
   2           7      15