Rename a single pandas DataFrame column without knowing column name

Question

I know I can rename single pandas.DataFrame columns with:

drugInfo.rename(columns = {'col_1': 'col_1_new_name'}, inplace = True)

But I'd like to rename a column based on its index (without knowing its name) - although I know dictionaries don't have it). I would like to rename column number 1 like this:

drugInfo.rename(columns = {1: 'col_1_new_name'}, inplace = True)

But in the DataFrame.columns dict there is no '1' entry, so no renaming is done. How could I achieve this?

Dreadful

Answer 1

Should work:

drugInfo.rename(columns = {list(drugInfo)[1]: 'col_1_new_name'}, inplace = True)

Example:

In [18]:

df = pd.DataFrame({'a':randn(5), 'b':randn(5), 'c':randn(5)})
df
Out[18]:
          a         b         c
0 -1.429509 -0.652116  0.515545
1  0.563148 -0.536554 -1.316155
2  1.310768 -3.041681 -0.704776
3 -1.403204  1.083727 -0.117787
4 -0.040952  0.108155 -0.092292
In [19]:

df.rename(columns={list(df)[1]:'col1_new_name'}, inplace=True)
df
Out[19]:
          a  col1_new_name         c
0 -1.429509      -0.652116  0.515545
1  0.563148      -0.536554 -1.316155
2  1.310768      -3.041681 -0.704776
3 -1.403204       1.083727 -0.117787
4 -0.040952       0.108155 -0.092292

It is probably more readable to index into the dataframe columns attribute:

df.rename(columns={df.columns[1]:'col1_new_name'}, inplace=True)

So for you:

drugInfo.rename(columns = {drugInfo.columns[1]: 'col_1_new_name'}, inplace = True)

Zoezoeller

Answer 2

To change a column name by index, one could alter the underlying array of df.columns by index. So

df.columns.array[1] = 'col_1_new_name'
# or
df.columns.values[1] = 'col_1_new_name'
# or 
df.columns.to_numpy()[1] = 'col_1_new_name'

They all perform the following transformation (without referencing B, it is changed):

However, if a new dataframe copy needs to be returned, rename method is the way to go (as suggested by EdChum):

df1 = df.rename(columns={list(df)[1]: 'col_1_new_name'})

If df has many columns, instead of list(df), it might be worth it to call islice() from the standard itertools library to efficiently select a column label (e.g. the second column name):

from itertools import islice
df1 = df.rename(columns={next(islice(df, 1, 2)): 'col_1_new_name'})

Symbolism