How to convert a pandas dataframe to numeric future proof?

Question

Until now I used to convert all values in a pandas dataframe with combined numerical and string entries to numeric values if possible in one easy step, using .map and .to_numeric with "errors = 'ignore'".

It worked perfectly, but after updating to the latest version of Pandas (2.2.3) I get a FutureWarning.

import pandas as pd
A = pd.DataFrame({
    'x' : ['1','2','3'],
    'y' : ['not_a_number','5',9999],
    }) # example data
B = A.map(pd.to_numeric, errors = 'ignore')

# FutureWarning: errors='ignore' is deprecated and will raise in a future version. Use to_numeric without passing errors and catch exceptions explicitly instead B = A.map(pd.to_numeric, errors = 'ignore')

How could I code this future proof in an elegant, vectorised way?

I could not think of any solution that is not very cumbersome (looping over each individual entry of the dataframe).

Answer 1

When you use errors='ignore', to_numeric returns the original Series.

As mentioned in the documentation:

errors {‘ignore’, ‘raise’, ‘coerce’}, default ‘raise’

• If ‘raise’, then invalid parsing will raise an exception.
• If ‘coerce’, then invalid parsing will be set as NaN.
• If ‘ignore’, then invalid parsing will return the input.

Changed in version 2.2. “ignore” is deprecated. Catch exceptions explicitly instead.

Catch the error explicitly if you want to keep the previous behavior:

def to_numeric(s):
    try:
        return pd.to_numeric(s, errors='raise')
    except ValueError:
        return s
    
A.apply(to_numeric)

NB. use apply rather than map for a vectorial operation.

Relevant issues: #54467, #59221.