Select by first index of multiindex (Pandas DataFrame)

Question

I have a DataFrame with index (a, b)

df = pd.DataFrame({'a' : [1, 2, 3, 1], 'b' : [11, 11, 11, -7], 'val' : [10, 19, 24, 12]})
df.set_index(['a', 'b'], inplace=True)

a | b  | val
1 | 11 | 10
2 | 11 | 19
3 | 11 | 24
1 | -7 | 12

I'd like to pick from it only the rows with b == 11.

I've tried

df[df.index[1] == 11]

KeyError: False

It failed. Could you help?

That's the solution) @shubham-sharma, could you post it so I can accept it? — GooseIt
– GooseIt, Commented Jul 4, 2022 at 17:38
@Gooselt I think this question might be a duplicate so I'll pass on adding an answer but anyways thanks! — Shubham Sharma
– Shubham Sharma, Commented Jul 4, 2022 at 17:46
There is an alternative with loc to keep the second level. — Corralien
– Corralien, Commented Jul 4, 2022 at 18:15

Corralien · Accepted Answer · 2022-07-04 18:14:09Z

2

There are 2 options to pick the rows with b==11.

>>> df.xs(11, level=1)
   val
a     
1   10
2   19
3   24

>>> df.loc[pd.IndexSlice[:, 11], :]
      val
a b      
1 11   10
2 11   19
3 11   24

answered Jul 4, 2022 at 18:14

Corralien

121k8 gold badges44 silver badges69 bronze badges

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

1 Comment

The IndexSlice tip is to me the most expressive option. Most familiar in its use too.