How to parse different string date formats?

问题

Working on a table with a mix of different strings where it's possible to derive a date.

    period
0   Q2 '20 Base
1   Q3 '20 Base
2   Q1 '21 Base
3   February '20 Base
4   March '20 Peak
5   Summer 22 Base
6   Winter 20 Peak
7   Summer 21 Base
8   Year 2021
9   October '21 Peak

I'd like to be able to parse this into a timestamp for analysis in python. First, ideally I want to parse into 4 new columns 1) day 2) month 3) quarter 4) year. Then use these columns to make a datetime (DD-MM-YYYY).

    period             day  month quarter year
0   Q2 '20 Base         01  04    1       2020
1   Q3 '20 Peak         01  07    3       2020
2   Q1 '21 Base         01  01    1       2021
3   February '20 Base   01  02    1       2020
4   March '20 Peak      01  03    1       2020
5   Summer 22 Base      01  04    2       2022
6   Winter 20 Peak      01  10    4       2020
7   Summer 21 Base      01  04    2       2021
8   Year 2021           01  01    1       2021
9   October '21 Base    01  10    4       2021

How can I parse this into the 4 new columns?

回答1:

My idea is to set up a dictionary data structure for your identifiers like this:

datemap = { 'January' :  {'day' : 1, 'month' : 1, 'quarter' : 1}, 
            'February' : {'day' : 1, 'month' : 2, 'quarter' : 1}, 
            'March' :    {'day' : 1, 'month' : 3, 'quarter' : 1}, 
            # and so on ...
            'Spring' : {'day' : 1, 'month' : 1, 'quarter' : 1}, 
            'Summer' : {'day' : 1, 'month' : 4, 'quarter' : 2}, 
            'Fall' :   {'day' : 1, 'month' : 7, 'quarter' : 3}, 
            'Winter' : {'day' : 1, 'month' : 10, 'quarter' : 4}, 
            'Q1' : {'day' : 1, 'month' : 1, 'quarter' : 1}, 
            'Q2' : {'day' : 1, 'month' : 4, 'quarter' : 2}, 
            'Q3' : {'day' : 1, 'month' : 7, 'quarter' : 3}, 
            'Q4' : {'day' : 1, 'month' : 10, 'quarter' : 4}, 
            'Year' : {'day' : 1, 'month' : 1, 'quarter' : 1} }

Then you can transform a given value r['period'] by looking at the first word r['period'].split()[0] (or second word for the year) like this:

df['day'] = df.apply (lambda r: datemap[r['period'].split()[0]]['day'], axis=1)
df['month'] = df.apply (lambda r: datemap[r['period'].split()[0]]['month'], axis=1)
df['quarter'] = df.apply (lambda r: datemap[r['period'].split()[0]]['quarter'], axis=1)
df['year'] = df.apply (lambda r: "20" + r['period'].split()[1][-2:], axis=1)

来源：https://stackoverflow.com/questions/59705281/how-to-parse-different-string-date-formats

标签

python

pandas

datetime-format