根据 DataFrame 中的值重命名具有相同列名的列

Question

I have a DataFrame which can contain columns with the same column name. Based on the value I want to rename the column name so there are no duplicates. I've tried a few things, but every time I try to iterate over the columns and rename them I end up with the column name. df.rename(columns=df.columns[i]: 'some_name'}) seems to use the column name as well.

Let's say I have a dataframe;

df = pd.DataFrame({"A": [10kg], "B": [4], "A": [4%]})

I would like to rename the column(s) named "A" based on the row value so that I get

   A     B  A%
0  10kg  4  4

I tried something like this:

for i in range(0, len(df.columns)):
    if 'A' in df.columns[i]:
        if '%' in df.iloc[:,i].values[0]:
            df = df.rename(columns={df.columns[i]: 'A_%'})

But this also renames the first column 'A'. Is there another way to rename it based on location?

Answer 1

Single list comprehension for new column names:

import pandas as pd

df = pd.concat([pd.DataFrame({"A": ['10kg'], "B": ['4']}), 
                pd.DataFrame({"A": ['4%']})], axis=1)


df.columns = [c + '_%'
              if df.applymap(lambda x: '%' in x).any(axis=0).iloc[ic]
              else c for ic, c in enumerate(df.columns)]

Edit -- better:

import pandas as pd

df = pd.concat([pd.DataFrame({"A": ['10kg'], "B": ['4']}), 
                pd.DataFrame({"A": ['4%']})], axis=1)

has_percentage = df.applymap(lambda x: '%' in x).any(axis=0)
df.columns = [c + '_%' if has_percentage.iloc[ic]
              else c for ic, c in enumerate(df.columns)]

Answer 2

You could create a list with all the column names, change the i'th column name in that list and use that list to redefine the column names:

for i in range(0, len(df.columns)):
    if 'A' in df.columns[i]:
        if '%' in df.iloc[:,i].values[0]:
            columnnames = list(df.columns)
            columnnames[i] = 'A_%'
            df.columns = columnnames