Sort pandas dataframe on string date without conversion [duplicate]

Question

Set-up

I have a dataframe with amongst others a column with dates in string.

The date format is day/month/year, e.g. 05/12/2018.

The test_df dataframe looks like,

   date_created
0    05/12/2018
1    04/12/2018
2    04/12/2018
3    03/12/2018
4    02/12/2018
5    30/11/2018
6    30/11/2018
7    30/11/2018
8    30/11/2018
9    29/11/2018
10   02/12/2018
11   29/11/2018
12   29/11/2018

---

Problem

I want this to be sorted on date, such that I have,

   date_created
0    05/12/2018    
1    04/12/2018
2    04/12/2018
3    03/12/2018
10   02/12/2018    
4    02/12/2018
5    30/11/2018
6    30/11/2018
7    30/11/2018
8    30/11/2018
9    29/11/2018
11   29/11/2018
12   29/11/2018

but following this example,

test_df.sort_values(by='date_created', inplace=False, ascending=True)

gives,

   date_created
4    02/12/2018
10   02/12/2018
3    03/12/2018
1    04/12/2018
2    04/12/2018
0    05/12/2018
9    29/11/2018
11   29/11/2018
12   29/11/2018
5    30/11/2018
6    30/11/2018
7    30/11/2018
8    30/11/2018

How do I reverse the order within the month?

Answer 1

The recommended solution is to convert your series to datetime, then sort:

df['date_created'] = pd.to_datetime(df['date_created'], dayfirst=True)

res = df.sort_values('date_created', ascending=False)

print(res)

   date_created
0    2018-12-05
1    2018-12-04
2    2018-12-04
3    2018-12-03
4    2018-12-02
10   2018-12-02
5    2018-11-30
6    2018-11-30
7    2018-11-30
8    2018-11-30
9    2018-11-29
11   2018-11-29
12   2018-11-29

If you insist on keeping an object dtype series, you can then convert back to strings in a separate operation:

res['date_created'] = res['date_created'].dt.strftime('%d/%m/%Y')