CODEWITHSUNDEEP
pythonpandasdataframe
Venkatesh Malhotra
Venkatesh Malhotra

Reputation: 251

Changing values in a dataframe based on condition

Here is the dataframe which I have:

import pandas as pd
list = ['RK','G', 'CUSTOM', 'CUSTOM', 'KL', 'kj']
df=pd.DataFrame(list,columns=['A'])

What I would like to do here is to find out all the entries in column A which are equal to CUSTOM and replace it with CUSTOM1 AND CUSTOM2.

The output should be:

       A
0      RK
1      G
2    CUSTOM1
3    CUSTOM2
4     KL
5     kj

Had there been 3 CUSTOM instead of 2, the output should be CUSTOM1 , CUSTOM2 and CUSTOM3.

Thanks.

Upvotes: 2

Views: 153

Answers (4)

juanpa.arrivillaga
juanpa.arrivillaga

Reputation: 96286

I'm interested to see how this approach will work compared to the others:

In [10]: import itertools

In [11]: counter = itertools.count(1)

In [12]: df.A.str.replace(r"CUSTOM", lambda x: f"CUSTOM{next(counter)}")
Out[12]:
0         RK
1          G
2    CUSTOM1
3    CUSTOM2
4         KL
5         kj
Name: A, dtype: object

Note, you need pandas version >= 0.20.0 for pd.Series.str.replace to accept a callable.

If you don't have Python 3.6 f-strings, you can also just do:

In [14]: counter = itertools.count(1)

In [15]: df.A.str.replace(r"CUSTOM", lambda x: "CUSTOM{}".format(next(counter)))
Out[15]:
0         RK
1          G
2    CUSTOM1
3    CUSTOM2
4         KL
5         kj
Name: A, dtype: object

Upvotes: 2

cs95
cs95

Reputation: 403128

Using cumsum + np.where - 

m = df.A.eq('CUSTOM')
df.A = np.where(m, df.A + m.cumsum().astype(str), df.A)

df
         A
0       RK
1        G
2  CUSTOM1
3  CUSTOM2
4       KL
5       kj

A similar solution using pd.Series.where/mask - 

df.A = df.A.where(~m, df.A + m.cumsum().astype(str))

Or,

df.A = df.A.mask(m, df.A + m.cumsum().astype(str))


df
         A
0       RK
1        G
2  CUSTOM1
3  CUSTOM2
4       KL
5       kj

Upvotes: 2

kjmerf
kjmerf

Reputation: 4345

You could do it with a loop:

count = 1
for index, row in df.iterrows():
    if row['A'] == 'CUSTOM':
        row['A'] += str(count)
        count += 1

print(df)

#output
         A
0       RK
1        G
2  CUSTOM1
3  CUSTOM2
4       KL
5       kj

Upvotes: 1

BENY
BENY

Reputation: 323376

By using cumcount + count 

df.A=df.A+(df.groupby('A').cumcount()+1).where(df.groupby('A').A.transform('count').ne(1),'').astype(str)
df.A
Out[208]: 
0         RK
1          G
2    CUSTOM1
3    CUSTOM2
4         KL
5         kj
dtype: object

Upvotes: 2

Related Questions