Pandas Series.rename未反映在DataFrame列中

Question

I'm trying to rename a column by validating the values in the particular columns. Here is the set-up:

In [9]: import pandas as pd

In [10]: df = pd.DataFrame(
    ...:         {"unknown_field": ['bob@gmail.com', 'shirley@gmail.com', 'groza@pubg.com']}
    ...:     )

In [11]: df
Out[11]: 
       unknown_field
0      bob@gmail.com
1  shirley@gmail.com
2     groza@pubg.com

Using a validate_column(ser) , which takes Pandas.Series object as parameter, it validates the values in that column and modifies the column name of that particular column with a pre-defined set of column names. To make it simple, in this example, the column is validated as an email column.

In [12]: def validate_column(ser):
    ...:     # Value validation method returns that this column is email column
    ...:     ser.rename('email', inplace=True)
    ...:     

The current name of the unknown_field is: unknown_field , and as expected, the name changes to email after executing validate_column method:

In [13]: df.unknown_field
Out[13]: 
0        bob@gmail.com
1    shirley@gmail.com
2       groza@pubg.com
Name: unknown_field, dtype: object

In [14]: validate_column(df.unknown_field)

In [15]: df.unknown_field
Out[15]: 
0        bob@gmail.com
1    shirley@gmail.com
2       groza@pubg.com
Name: email, dtype: object

However, the column names within the df is not modified as I expected. It still named as unknown_field within the df variable:

In [16]: df
Out[16]: 
       unknown_field
0      bob@gmail.com
1  shirley@gmail.com
2     groza@pubg.com

Currently, I use the following code to manually modify the column name within my df variable.

In [17]: for col in df.select_dtypes(object):
    ...:     df.rename(columns={col: df[col].name}, inplace=True)
    ...:     

In [18]: df
Out[18]: 
               email
0      bob@gmail.com
1  shirley@gmail.com
2     groza@pubg.com

In [19]: 

My question is:

Is there a more efficient/straightforward method for renaming the Series and directly reflected in the DataFrame ?

Answer 1

Re-write your function to accept two parameters:

def validate_column(df, col_name):
    # Value validation method returns that this column is email column
    return df.rename({col_name : 'email'}, axis=1)

Now, call your function through DataFrame.pipe :

df.pipe(validate_column, col_name='unknown_field')

               email
0      bob@gmail.com
1  shirley@gmail.com
2     groza@pubg.com

Very clean. This is useful if you want to chain validations:

(df.pipe(validate_column, col_name='unknown_field')
   .pipe(validate_column, col_name='some_other_field')
   .pipe(validate_column, col_name='third_field')
)

... or modify validate_column to validate multiple columns at a time.

Note that the renaming is no longer done in-place, and whatever result is returned from pipe needs to be assigned back.

Answer 2

Use dataframe's rename function and set columns argument.

import pandas as pd
df = pd.DataFrame({"unknown_field": ['bob@gmail.com', 'shirley@gmail.com', 'groza@pubg.com']})
df = df.rename(columns={'unknown_field': 'email'})

Output:

    email
0   bob@gmail.com
1   shirley@gmail.com
2   groza@pubg.com