CODEWITHSUNDEEP
pythonpandasfilter
lolo
lolo

Reputation: 646

Group in Python Pandas based on a date filter

I have the following dataframe. I need to group by ID adding the PROFIT column based on the last 30 and 60 days.

import pandas as pd
df = pd.DataFrame({"ID":[1,1,1,1,2,2,2,2],"DATE":['2019-04-03','2019-03-03','2019-03-01','2019-02-03','2019-02-01','2019-01-01','2019-01-06','2019-04-03'],"PROFIT":[10,20,30,60,90,100,20,10]})

    ID  DATE    PROFIT
0   1   2019-04-03  10
1   1   2019-03-03  20
2   1   2019-03-01  30
3   1   2019-02-03  60
4   2   2019-02-01  90
5   2   2019-01-01  100
6   2   2019-01-06  20
7   2   2019-04-03  10

Final Result:

df_end = pd.DataFrame({"ID":[1,1,2,2],"TIME":[30,60,30,60],"SUM_PROFIT":[10,60,10,90]})

    ID  TIME    SUM_PROFIT
0   1   30      10
1   1   60      60
2   2   30      10
3   2   60      90

Upvotes: 0

Views: 86

Answers (1)

Scott Boston
Scott Boston

Reputation: 153460

IIUC, then you can try something like this:

timespan = [30, 60]
pd.concat([df.sort_values('DATE', ascending=False)
             .groupby(['ID'])
             .apply(lambda x: x.loc[x['DATE'].head(1).values[0]-x['DATE']<=pd.Timedelta(days=t),'PROFIT'].sum())
             .rename('SUM_PROFIT').reset_index().assign(TIME = t) for t in timespan],
          ignore_index=True)

Output:

   ID  SUM_PROFIT  TIME
0   1          10    30
1   2          10    30
2   1         120    60
3   2          10    60

Upvotes: 1

Related Questions