How can I remove all non-numeric characters from all the values in a particular column in pandas dataframe?

user_3436837 0
Viewed 64192

I have a dataframe which looks like this:

     A       B           C
1   red78   square    big235
2   green   circle    small123
3   blue45  triangle  big657

I need to be able to remove the non-numeric characters from all the rows in column C so that my dataframe looks like:

     A       B           C
1   red78   square    235
2   green   circle    123
3   blue45  triangle  657

I tried using the following but get the error expected string or buffer:

import re
dfOutput.imgID = dfOutput.imgID.apply(re.sub('[^0-9]','', dfOutput.imgID), axis = 0)

What should I do instead?

Code to create dataframe:

dfObject = pd.DataFrame()
dfObject.set_value(1, 'A', 'red78')
dfObject.set_value(1, 'B', 'square')
dfObject.set_value(1, 'C', 'big235')
dfObject.set_value(2, 'A', 'green')
dfObject.set_value(2, 'B', 'circle')
dfObject.set_value(2, 'C', 'small123')
dfObject.set_value(3, 'A', 'blue45')
dfObject.set_value(3, 'B', 'triangle')
dfObject.set_value(3, 'C', 'big657')
pandas dataframe python
6 Answers

To remove all non-digit characters from strings in a Pandas column you should use str.replace with \D+ or [^0-9]+ patterns:

dfObject['C'] = dfObject['C'].str.replace(r'\D+', '')

Or, since in Python 3, \D is fully Unicode-aware by default and thus does not match non-ASCII digits (like ۱۲۳۴۵۶۷۸۹, see proof) you should consider

dfObject['C'] = dfObject['C'].str.replace(r'[^0-9]+', '')

So,

import re
print ( re.sub( r'\D+', '', '1۱۲۳۴۵۶۷۸۹0') )         # => 1۱۲۳۴۵۶۷۸۹0
print ( re.sub( r'[^0-9]+', '', '1۱۲۳۴۵۶۷۸۹0') )     # => 10
user_38329700

After 2 years, to help others, I actually think that you were very close to the answer. I have used your logic but made it work. basically you create a function that does the clean up and then apply it to the column C.

import pandas as pd
import re

df = pd.DataFrame({
     'A': ['red78', 'green', 'blue45'],
     'B': ['square', 'circle', 'triangle'],
    'C': ['big235', 'small123',  'big657']
})

def remove_chars(s):
    return re.sub('[^0-9]+', '', s) 

df['C'] = df['C'].apply(remove_chars)
df

Result below:

A   B   C
0   red78   square  235
1   green   circle  123
2   blue45  triangle    657
user_101002720

Suppose we want to remove all characters from all the columns in one.

import pandas as pd
 

df = pd.DataFrame({
     'A': ['red78', 'green', 'blue45'],
     'B': ['square', 'circle', 'triangle'],
    'C': ['big235', 'small123',  'big657']
})
 df[['A', 'C']] = df[['A', 'C']].replace(regex=[r'\D+'], value="") 

Output:
   A         B    C
0  78    square  235
1        circle
2  45  triangle  657

user_104847800
Related