根据模式替换某些值并在 pandas 中提取 substring

Question

Pandas Dataframe with col1 that contains various dates

 col1
 Q2 '20
 Q1 '21
 May '20
 June '20
 25/05/2020
 Q4 '20+Q1 '21
 Q2 '21+Q3 '21
 Q4 '21+Q1 '22

I want to replace certain values in col1 that match a pattern. For the values that contain 2 quarters with "+" I want to return a season in string plus the first year contained in the pattern. I want to leave the other values as they are.

For example:

1) Q4 '20+Q1 '21 should be 'Winter 20'

2) Q2 '21+Q3 '21 should be 'Summer 21'

3) Q4 '21+Q1 '22 should be 'Winter 21'

Desired output:

col1
Q2 '20
Q1 '21
May '20
June '20
25/05/2020
Winter 20
Summer 20
Winter 21

I have tried with a few methods like replace, split, extract. But I am not solving the problem. Using a dictionary would not be helpful because the df is quite big with lots variants of the Q4 'XX+Q1 'XX and Q2 'XX +Q3 'XX

Answer 1

You could do it matching multiple patterns one for each season:

df = pd.DataFrame({'col1': [
"Q2 '20",
"Q1 '21",
"May '20",
"June '20",
"25/05/2020",
"Q4 '20+Q1 '21",
"Q2 '21+Q3 '21",
"Q4 '21+Q1 '22"]})

seasons = {
r"Q4 '(\d*)\+Q1 .*": r'Winter \1',
r"Q1 '(\d*)\+Q2 .*": r'Spring \1',
r"Q2 '(\d*)\+Q3 .*": r'Summer \1',
r"Q3 '(\d*)\+Q4 .*": r'Autumn \1'
}

df.col1.replace(seasons, regex=True)

0        Q2 '20
1        Q1 '21
2       May '20
3      June '20
4    25/05/2020
5     Winter 20
6     Summer 21
7     Winter 21

Or the other version which I think is more efficient because I am matching only one regex but i use global variables so i am not sure which version is better.

seasons = {
'Q4Q1': 'Winter',
'Q1Q2': 'Spring',
'Q2Q3': 'Summer',
'Q3Q4': 'Autumn'
}
pattern = re.compile(r"(Q\d) '(\d*)\+(Q\d) .*")

def change_to_season(row):
    match = pattern.match(row)
    if match:
        season = seasons[match.group(1) + match.group(3)]
        year = match.group(2)
        return season + ' ' + year
    else:
        return row

df.col1.apply(change_to_season)

Answer 2

'''
col1
Q2 '20
Q1 '21
May '20
June '20
25/05/2020
Q4 '20+Q1 '21
Q2 '21+Q3 '21
Q4 '21+Q1 '22
'''

import pandas as pd

df = pd.read_clipboard(sep="!")

print(df)

Output:

           col1
0         Q2 '20
1         Q1 '21
2        May '20
3       June '20
4     25/05/2020
5  Q4 '20+Q1 '21
6  Q2 '21+Q3 '21
7  Q4 '21+Q1 '22

.

import re 

def regex_filter(val):
    regex = re.compile(r"([Q][1-4])+ '(\d+)\+([Q][1-4])+ '(\d+)")
    result = regex.split(val)
    result = [val for val in result if val]
    if 'Q3' in result:
        result = 'Summer '+result[-1]
    elif 'Q1' in result:
        result = 'Winter '+result[1]
    else:
        result = ''.join(result)

    return result

df['col1'] = df['col1'].apply(regex_filter)



print(df)

Output:

         col1
0      Q2 '20
1      Q1 '21
2     May '20
3    June '20
4  25/05/2020
5   Winter 20
6   Summer 21
7   Summer 21