python - 在 for 循环中对 pandas groupby 对象进行操作

Question

我有一个像

这样的数据框

df = pd.DataFrame({'A': ['foo', 'bar','cat'] * 3,
             'B': ['i','who','he','am','are','said','satya','you','hello'],
             'C': [5, 2, 3, 4, 6, 9,12,23,45]})
df
Out[47]: 
A      B   C
0  foo      i   5
1  bar    who   2
2  cat     he   3
3  foo     am   4
4  bar    are   6
5  cat   said   9
6  foo  satya  12
7  bar    you  23
8  cat  hello  45

我将它们分组如下:

g = df.groupby('A')
for name, group in g:
     print(name)
     print(group)

获取 o/p 为:

bar
     A    B   C
1  bar  who   2
4  bar  are   6
7  bar  you  23
cat
     A      B   C
2  cat     he   3
5  cat   said   9
8  cat  hello  45
foo
     A      B   C
0  foo      i   5
3  foo     am   4
6  foo  satya  12

所以我需要做的是我想遍历每个组并再次针对每个组我将遍历(我不知道)列并连接 B 列的值，以便我可以得到类似的东西

bar = who are you
cat = he said hello
foo = i am satya

我无法对其执行 agg 函数。谁能建议如何做到这一点？

Answer 1

IIUC 您可以在 A 上groupby，然后应用一个 lambda 来加入字符串值:

In [185]:
df.groupby('A')['B'].apply(' '.join)

Out[185]:
A
bar      who are you
cat    he said hello
foo       i am satya
Name: B, dtype: object