CODEWITHSUNDEEP
pythonpandasdataframesortingdata-analysis
Naveen Reddy Marthala
Naveen Reddy Marthala

Reputation: 3123

Pandas - how to sort week and year numbers formatted as strings?

I have a pandas dataframe like this, which sorted like:

>>> weekly_count.sort_values(by='date_in_weeks', inplace=True)
>>> weekly_count.loc[:9,:]

  date_in_weeks count
0   1-2013      362
1   1-2014      378
2   1-2015      201
3   1-2016      294
4   1-2017      300
5   1-2018      297
6   10-2013     329
7   10-2014     314
8   10-2015     324
9   10-2016     322

in above data, first column, all rows of date_in_weeks are simply "week number of a year - year". I now want to sort it like this:

  date_in_weeks count
0   1-2013      362
6   10-2013     329
1   1-2014      378
7   10-2014     314
2   1-2015      201
8   10-2015     324
3   1-2016      294
9   10-2016     322
4   1-2017      300
5   1-2018      297

How do i do this?

Upvotes: 2

Views: 1069

Answers (3)

jezrael
jezrael

Reputation: 863236

Use Series.argsort with converted to datetimes with format %W week number of the year, link:

df = df.iloc[pd.to_datetime(df['date_in_weeks'] + '-0',format='%W-%Y-%w').argsort()]
print (df)
  date_in_weeks  count
0        1-2013    362
6       10-2013    329
1        1-2014    378
7       10-2014    314
2        1-2015    201
8       10-2015    324
3        1-2016    294
9       10-2016    322
4        1-2017    300
5        1-2018    297

Upvotes: 5

anky
anky

Reputation: 75100

You can also convert to datetime , assign to the df, then sort the values and drop the extra col:

s = pd.to_datetime(df['date_in_weeks'],format='%M-%Y')
final = df.assign(dt=s).sort_values(['dt','count']).drop('dt',1)

print(final)

  date_in_weeks  count
0        1-2013    362
6       10-2013    329
1        1-2014    378
7       10-2014    314
2        1-2015    201
8       10-2015    324
3        1-2016    294
9       10-2016    322
4        1-2017    300
5        1-2018    297

Upvotes: 3

Celius Stingher
Celius Stingher

Reputation: 18377

You can try using auxiliary columns: 

import pandas as pd
df = pd.DataFrame({'date_in_weeks':['1-2013','1-2014','1-2015','10-2013','10-2014'],
                   'count':[362,378,201,329,314]})
df['aux'] = df['date_in_weeks'].str.split('-')
df['aux_2'] = df['aux'].str.get(1).astype(int)
df['aux'] = df['aux'].str.get(0).astype(int)
df = df.sort_values(['aux_2','aux'],ascending=True)
df = df.drop(columns=['aux','aux_2'])
print(df)

Output:

  date_in_weeks  count
0        1-2013    362
3       10-2013    329
1        1-2014    378
4       10-2014    314
2        1-2015    201

Upvotes: 2

Related Questions