df.replace call does not make changes to the original dataframe

Question

With pd.DataFrame.replace, I have been trying to figure out why the values that I replaced revert back to their original values when I replace another set in another column. How can you make the replacements permanent, below?

titanic.replace({'Pclass' : 
                {3 : 'Lower Class', 
                 2 : 'Middle Class', 
                 1: 'Upper Class'}})

You have to assign the result using titanic = titanic.replace(.... or use parameter inplace = True — Vaishali, Aug 31 '17 at 20:58

cs95 · Answer 1 · 2017-08-31T21:09:04.427

3

Read the df.replace documentation:

DataFrame.replace(to_replace=None, value=None, inplace=False, 
                       limit=None, regex=False, method='pad', axis=None)

Replace values given in to_replace with value.

Returns: filled : NDFrame

This isn't an in-place operation unless you specify it to be so.

titanic.replace({'Survived':(0:'False',1:'True')}, inplace=True)

or, assign the return value back.

titanic = titanic.replace({'Survived':(0:'False',1:'True')})

Don't try assigning when inplace=True because then, df.replace returns None.

edited Aug 31 '17 at 21:09

answered Aug 31 '17 at 20:59

cs95

379,657
97
704
746

thanks. i am very new to programming and it's hard for me to understand the documentation. your answer here helped a lot though. – techscolasticus Sep 02 '17 at 18:14
@techscolasticus If my answer helped, you can mark it accepted. Thank you. – cs95 Sep 02 '17 at 20:00
@coldspeed. I tried marking it as accepted but then it removes the mark from Daniel H's answer which was a bit easier for me to understand why my original method didn't work. – techscolasticus Sep 02 '17 at 20:04

score 3 · Answer 2 · answered Aug 31 '17 at 20:59

3

add inplace =True

titanic.replace({'Survived':(0:'False',1:'True')},inplace =True)

Also for kaggle titanic data, recommend map

d={0:'False',1:'True'}

titanic.Survived=titanic.Survived.map(d)

answered Aug 31 '17 at 20:59

BENY

317,841
20
164
234

what does map do? sorry, i'm very new to programming in general and have trouble understanding pandas documentation. – techscolasticus Sep 02 '17 at 18:12
@techscolasticus think about 'map' is 'merge', In your example , If value =0 after map it will return `False`, just like a dictionary , 0-->False – BENY Sep 02 '17 at 18:14

score 0 · Accepted Answer · answered Aug 31 '17 at 21:00

The code titanic.replace(...) creates a new DataFrame, and returns that. In order to have Python remember this DataFrame later instead of just getting rid of it, you need to assign it to something, with df = titatic.replace(...). Here df can be any name, including titanic; if you use the same name, it will get rid of the old DataFrame and use the new one from then on.

df.replace call does not make changes to the original dataframe

3 Answers3