Home > Back-end >  pandas: replace column value with keys and values in a dictionary of list values
pandas: replace column value with keys and values in a dictionary of list values

Time:05-10

I have a dataframe and a dictionary as follows (but much bigger),

import pandas as pd
df = pd.DataFrame({'text': ['can you open the door?','shall you write the address?']})

dic = {'Should': ['can','could'], 'Could': ['shall'], 'Would': ['will']}

I would like to replace the words in the text column if they can be found in dic list of values, so i did the following and it works for the lists that have one value but not for the other list,

for key, val in dic.items():
    if df['text'].str.lower().str.split().map(lambda x: x[0]).str.contains('|'.join(val)).any():
       df['text'] = df['text'].str.replace('|'.join(val), key, regex=False)
print(df)

my desired output would be,

              text
0   Should you open the door?
1  Could you write the address?

CodePudding user response:

The best is to change the logic and try to minimize the pandas steps.

You can craft a dictionary that will directly contain your ideal output:

dic2 = {v:k for k,l in dic.items() for v in l}
# {'can': 'Should', 'could': 'Should', 'shall': 'Could', 'will': 'Would'}

# or if not yet formatted:
# dic2 = {v.lower():k.capitalize() for k,l in dic.items() for v in l}

import re
regex = '|'.join(map(re.escape, dic2))

df['text'] = df['text'].str.replace(f'\b({regex})\b',
                                    lambda m: dic2.get(m.group()),
                                    case=False, # only if case doesn't matter
                                    regex=True)

output (as text2 column for clarity):

                           text                         text2
0        can you open the door?     Should you open the door?
1  shall you write the address?  Could you write the address?

CodePudding user response:

You can use lowercase in flatten dictionary to d for keys and values, then replace values with words boundaries and last use Series.str.capitalize:

d = {x.lower(): k.lower() for k, v in dic.items() for x in v}


regex = '|'.join(r"\b{}\b".format(x) for x in d.keys())
df['text'] = (df['text'].str.lower()
                        .str.replace(regex, lambda x: d[x.group()], regex=True)
                        .str.capitalize())
print(df)
                           text
0     Should you open the door?
1  Could you write the address?
  • Related