CODEWITHSUNDEEP
pythonstringpandas
the_reaper
the_reaper

Reputation: 83

grouping values of a dataframe with string

The following code works for classifying a single message

total_frame['dummy_message'][total_frame['Message'].str.contains(['rrc'],case = False)] = 'msg1' 

index  Message

0       rrc
1       as1
2       as1
3        a2
4       as1
5       a2

However if I want to classify all the messages in a message column, I want to use something like this

total_frame['dummy_message'[total_frame['Message'].str.contains(['rrc','as1','as2','a2'],case = False)] = 'msg1','msg2','msg3','msg4'

This doesn't work as str.contains doesnt accept a list. The output should look something like this

index  Message dummy message

0       rrc          msg1
1       as1          msg2
2       as1          msg2
3        a2          msg4
4       as2          msg3
5       a2           msg4

Is there any alternative?

Upvotes: 0

Views: 36

Answers (1)

cs95
cs95

Reputation: 402553

Initialise a mapping of substrings to categories, then use str.extract to extract, and map to classify them:

mapping = dict(zip(
    ['rrc', 'as1', 'as2', 'a2'], 
    ['msg1', 'msg2', 'msg3', 'msg4']))

df['category'] = (
    df['Message'].str.extract(r'(?i)({})'.format('|'.join(mapping)), expand=False)
                 .map(mapping))

If case-insensitivity is important, modify your regex as: r'(?i)({})'.format('|'.join(mapping)).

Minimal Code Example

df = pd.DataFrame({'Message': ['this is as1', 'abcd rrc', 'xyz as2']})
df

       Message
0  this is as1
1     abcd rrc
2      xyz as2

df['category'] = (
    df['Message'].str.extract(r'({})'.format('|'.join(mapping)), expand=False)
                 .map(mapping))
df

       Message category
0  this is as1     msg2
1     abcd rrc     msg1
2      xyz as2     msg3

Upvotes: 1

Related Questions