Pandas - remove the label of the column index

Question

I have a dataframe as follows:

PLEASE_REMOVE  2013  2014  2015
 THIS_IS_EASY
-------------------------------
          Bob     0     3     4
         Mary     2     3     6

The years (2013, 2014, 2015) are the column index labels. The names (Mary, Bob) are the row index labels.

I have somehow managed to get labels for the row indices and column indices.

Using df.index.names = [''] I can remove the THIS_IS_EASY bit.

How can I remove the PLEASE_REMOVE bit?

Desired output is:

               2013  2014  2015
 -------------------------------
          Bob     0     3     4
         Mary     2     3     6

What's your data source and sample code? – misterflister Oct 28 '18 at 23:21 — misterflister, Oct 28 '18 at 23:21

Peter Leimbigler · Accepted Answer · 2020-08-01T15:52:39.150

37

New answer for pandas 1.x, submitted by Ian Logie

df.columns.name = None

Simply delete the name of the columns:

del df.columns.name

Also, note that df.index.names = [''] is not quite the same as del df.index.name.

edited Aug 01 '20 at 15:52

answered Oct 28 '18 at 23:24

Peter Leimbigler

5

This worked, and I switched to the `del df.index.name` which I realise now gave me a more correct solution. – user9588528 Oct 28 '18 at 23:33
does this work on multipindex? I get 'AttributeError: name' – user-2147482637 Aug 02 '19 at 06:40
@user-2147482637, use MultiIndex.rename() for that. For example, if df has 2 index levels, you can rename the outermost level (level=0) with: `df.index = df.index.rename('new_level_name', level=0)` – Peter Leimbigler Aug 03 '19 at 14:19
2

This seems to have broken in Pandas 1.0.3, [this is now the best answer](https://stackoverflow.com/a/61312610/2098573). – rysqui Jul 01 '20 at 20:40
If you do this still "none" or "NAN" will appear while you visualize your dataframe let's say for example in a jupyter notebook. I've found out from [this solution](https://stackoverflow.com/a/57738153/5127304) that actually the best way is to put `df.columns.name=''` – Elias Jan 22 '21 at 07:53
1

In 1.2.4, df.columns.name throws error. It is df.columns.names and it does not take single value, it expects list like df.columns.names = [None, None] if you have multi header. and to remove index label you have to use temp_df.rename_axis(None, inplace=True) seperatly. – shiv Apr 21 '21 at 11:02

Rocky Li · Answer 2 · 2018-10-28T23:40:06.683

18

Like this:

df = df.rename_axis(None)

edited Oct 28 '18 at 23:40

answered Oct 28 '18 at 23:26

Rocky Li

1

Not everything. It will only get rid of "THIS IS EASY" in the example in the original post because of the default. `df = df.rename_axis(None, axis=1)` or `df = df.rename_axis(None, axis="columns")` to remove the overall name of the columns, i.e. "PLEASE REMOVE" in the example in the original post. – Alper Aug 21 '22 at 01:54

Ian Logie · Answer 3 · 2020-04-20T02:57:23.053

16

In Pandas 1.0.3 the following works:

df.columns.name = None

The 'accepted answer' above, del df.columns.name, leads to: 'AttributeError: can't delete attribute'

edited Apr 20 '20 at 02:57

answered Apr 19 '20 at 22:43

Ian Logie

2

In pandas 1.0.3, the accepted answer does not work anymore. @Ian Logie you are right. – Eiffelbear May 03 '20 at 13:28
Edited my answer (which was accepted back in October 2018) to reflect this correct way to remove the name of the column index. Thanks @IanLogie! – Peter Leimbigler Mar 31 '21 at 03:56

score 4 · Answer 4 · answered Jun 02 '20 at 06:26

4

In Pandas 1.0.3 I have used the following :

df.rename_axis(None, axis = 1)

answered Jun 02 '20 at 06:26

DBear

score 2 · Answer 5 · answered Sep 18 '20 at 02:57

2

columns.name will not work since poster want to remove the index name

df.index.name = None

answered Sep 18 '20 at 02:57

Michael Li