255

I was looking for an elegant way to change a specified column name in a DataFrame.

play data ...

import pandas as pd d = { 'one': [1, 2, 3, 4, 5], 'two': [9, 8, 7, 6, 5], 'three': ['a', 'b', 'c', 'd', 'e'] } df = pd.DataFrame(d)

The most elegant solution I have found so far ...

names = df.columns.tolist() names[names.index('two')] = 'new_name' df.columns = names

I was hoping for a simple one-liner ... this attempt failed ...

df.columns[df.columns.tolist().index('one')] = 'another_name'

Any hints gratefully received.

Improve this question
0

10 Answers 10

Reset to default
454

A one liner does exist:

In [27]: df=df.rename(columns = {'two':'new_name'}) In [28]: df Out[28]: one three new_name 0 1 a 9 1 2 b 8 2 3 c 7 3 4 d 6 4 5 e 5

Following is the docstring for the rename method.

 Definition: df.rename(self, index=None, columns=None, copy=True, inplace=False) Docstring: Alter index and / or columns using input function or functions. Function / dict values must be unique (1-to-1). Labels not contained in a dict / Series will be left as-is. Parameters ---------- index : dict-like or function, optional Transformation to apply to index values columns : dict-like or function, optional Transformation to apply to column values copy : boolean, default True Also copy underlying data inplace : boolean, default False Whether to return a new DataFrame. If True then value of copy is ignored. See also -------- Series.rename Returns ------- renamed : DataFrame (new object)
Improve this answer
Sign up to request clarification or add additional context in comments.

1 Comment

This does not work for me unless I use the inplace=True as shown in @Jeong-Yoon Lee reply.
135

Since inplace argument is available, you don't need to copy and assign the original data frame back to itself, but do as follows:

df.rename(columns={'two':'new_name'}, inplace=True)
Improve this answer

Comments

51

What about?

df.columns[2] = "new_name"
Improve this answer

6 Comments

actually that doesn't work if later you use the column name on other operations like in df['new_name']
this answer was useful for me to change a specific column to a new name. 1st column is index 0, 2nd column is index 1, and so on. nice solution.. and i am sure this will help more people.. as the other solutions require you to know and copy the original column names beforehand.... while this is quick and dirty method.. which has its own uses.
@MasterYogurt your comment is not correct. It is possible to perform df['new_name'] (and other pandas things) after changing variables as outlined above. Your comment may have been valid when it was originally posted.
That being said, using the rename methods is a better solution.
@JacobH I still run into this issue with Pandas v1.0.3
|
18

If you know which column # it is (first / second / nth) then this solution posted on a similar question works regardless of whether it is named or unnamed, and in one line: https://stackoverflow.com/a/26336314/4355695

df.rename(columns = {list(df)[1]:'new_name'}, inplace=True) # 1 is for second column (0,1,2..)
Improve this answer

2 Comments

This works only if there is not "nan" in the column names, otherwise it will change all of them.
The only solution for me - I had a list of lists converted to a DF with columns 0,1,..250, and I only needed to "name" the first three columns. This worked perfectly.
11

Pandas 0.21 now has an axis parameter

The rename method has gained an axis parameter to match most of the rest of the pandas API.

So, in addition to this:

df.rename(columns = {'two':'new_name'})

You can do:

df.rename({'two':'new_name'}, axis=1)

or

df.rename({'two':'new_name'}, axis='columns')
Improve this answer

2 Comments

df.rename({'two':'new_name'}, axis='columns') Throws TypeError: Cannot specify both 'axis' and any of 'index' or 'columns'.
@HereHere Make sure you are on pandas version 0.21. Do pd.__version__ to check your version
4

For renaming the columns here is the simple one which will work for both Default(0,1,2,etc;) and existing columns but not much useful for a larger data sets(having many columns).

For a larger data set we can slice the columns that we need and apply the below code:

df.columns = ['new_name','new_name1','old_name']
Improve this answer

Comments

4

pandas version 0.23.4

df.rename(index=str,columns={'old_name':'new_name'},inplace=True)

For the record:

omitting index=str will give error replace has an unexpected argument 'columns'

Improve this answer

1 Comment

This extends nicely for more than one column: if you have a list of the old_names and the new-names (one for one), then you can do: df.rename(columns=dict(zip(old_names, new_names)), inplace=True)
3

Following short code can help:

df3 = df3.rename(columns={c: c.replace(' ', '') for c in df3.columns})

Remove spaces from columns.

Improve this answer

1 Comment

I kept getting AttributeError: 'int' object has no attribute 'replace' could you expand on that.
2

Another option would be to simply copy & drop the column:

df = pd.DataFrame(d) df['new_name'] = df['two'] df = df.drop('two', axis=1) df.head()

After that you get the result:

 one three new_name 0 1 a 9 1 2 b 8 2 3 c 7 3 4 d 6 4 5 e 5
Improve this answer

1 Comment

This method will not help in case the order of the columns indexes is important. The new column will be created at the end.
0 

size = 10 df.rename(columns={df.columns[i]: someList[i] for i in range(size)}, inplace = True)

Improve this answer

Comments

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.