How to add an index label using pandas MultiIndex.insert to an existing MultiIndex?

Question

With below code I tried to insert a new row label (two name-levels) but it appears that MultiIndex.insert() does not work according to the printed output. User xyzjayne mentions here "MultiIndex is immutable". Then why is there MultiIndex.insert and/or how to use MultiIndex.insert properly in below case to get desired result?

NOT an option:

add ('D', '9') to labels tuple list before MultiIndexing. This labels is general, has a default list order and is used a few times elsewhere in different settings. And thus prior to the creation of a DataFrame.

Versions :pandas 0.25.1 py37ha925a31_0

Desired result:

row_labels MultiIndex([('A', '1'),
                       ('A', '2'),
                       ('B', '3'),
                       ('B', '4'),
                       ('C', '5'),
                       ('C', '6'),
                       ('D', '9'),   # inserted row label ('D', '9')
                       ('D', '7'),
                       ('D', '8')],
                       names=['group', 'subgroup'])

My code:

import pandas as pd

labels = {'A' : ['1', '2'],
          'B' : ['3', '4'],
          'C' : ['5', '6'],
          'D' : ['7', '8']}

tpl = []

for g, sg in labels.items():
  for s in sg:
    tpl.append((g, s))

row_labels = pd.MultiIndex.from_tuples(tpl, names=['group', 'subgroup'])

print ('row_labels', row_labels)

i = row_labels.get_locs(('D',))

print ('\nD is found at row(s): ', i)

# add row at position 6
row_labels.insert(6, (('D', '9')))

# check for added label
print ('\nrow_labels', row_labels)

My current output:

row_labels MultiIndex([('A', '1'),
                       ('A', '2'),
                       ('B', '3'),
                       ('B', '4'),
                       ('C', '5'),
                       ('C', '6'),
                       ('D', '7'),
                       ('D', '8')],
                       names=['group', 'subgroup'])

D is found at row(s):  [6 7]

row_labels MultiIndex([('A', '1'),
                       ('A', '2'),
                       ('B', '3'),
                       ('B', '4'),
                       ('C', '5'),
                       ('C', '6'),
                       ('D', '7'),
                       ('D', '8')],
                       names=['group', 'subgroup'])

jezrael · Accepted Answer · 2020-02-04 14:38:43Z

1

You can assign back, Index.insert not working inplace:

print (pd.__version__)
0.25.1

# add row at position 6
row_labels = row_labels.insert(6, (('D', '9')))

# check for added label
print ('\nrow_labels', row_labels)
row_labels MultiIndex([('A', '1'),
            ('A', '2'),
            ('B', '3'),
            ('B', '4'),
            ('C', '5'),
            ('C', '6'),
            ('D', '9'),
            ('D', '7'),
            ('D', '8')],
           names=['group', 'subgroup'])

answered Feb 4, 2020 at 14:38

jezrael

868k103 gold badges1.4k silver badges1.3k bronze badges

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

3 Comments

ZF007 Over a year ago

.. I just figured # add row at position 6 row_labels = pd.MultiIndex.insert(row_labels, 6, ('D', '9')) out... but still in docs pandas 0.19 it mentions MultiIndex (loc, items).

jezrael Over a year ago

@ZF007 - My opinion - this function is used very rare, so maybe buggy. Obviously is MultiIndex converted to list of tuples or to DataFrames, procesing and then convert back to Multiindex.

ZF007 Over a year ago

Yeah.. buggy indeed. Thanks for jumping in. It saves me an hour or two of headachs on this.

ZF007 · Accepted Answer · 2020-02-04 14:54:05Z

0

As you can see from the discussion with jezrael the .insert appears to be sloppy made or missing text at the docs.

My alternative also works:

row_labels = pd.MultiIndex.insert(row_labels, 6, ('D', '9'))

And is not as such described here at the pandas docs 0.25.0 .

Here at 0.19.0 it is and the text should be:

MultiIndex.insert(index, loc, item)

and thus missing the previous made "MultiIndex" as "Index".

answered Feb 4, 2020 at 14:54

ZF007

3,7419 gold badges33 silver badges52 bronze badges

Collectives™ on Stack Overflow

How to add an index label using pandas MultiIndex.insert to an existing MultiIndex?

2 Answers 2

3 Comments

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

3 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related