根据列值提取 Pandas Dataframe 中的下一行

Question

Suppose I have the following Dataframe

  final  raw  act    wc             Start            Finish
   abc  xyz   30    M5  17-01-2022 06:00  14-07-2031 02:36
   abc  xyz   40    F4  17-01-2022 06:00  14-07-2031 02:36
   abc  xyz   50    F6  17-01-2022 06:00  14-07-2031 02:36
   abc  xyz   60    F8  17-01-2022 06:00  14-07-2031 02:36
   abc  pqr   40  M14S  17-01-2022 06:00  18-01-2026 17:21
   abc  pqr   50   M12  17-01-2022 06:00  18-01-2026 17:21
   abc  pqr   60  M14S  17-01-2022 06:00  18-01-2026 17:21
   abc  pqr   20    F3  17-01-2022 06:00  14-07-2031 02:36
   abc  pqr   40    F4  17-01-2022 06:00  14-07-2031 02:36
   abc  pqr   50    F6  17-01-2022 06:00  14-07-2031 02:36

I would like to take the two rows from here, one is

 abc  xyz   50  F6  17-01-2022 06:00  14-07-2031 02:36

another one is

abc  pqr   50    F6  17-01-2022 06:00  14-07-2031 02:36

The logic would be that for each raw , pick up the next row where wc is either F3, F4 and the act is maximum. Here for xyz , F4 is there, so the next row also for pqr , both F3, F4 are there but maximum act is 40.

I did it using pd.shift()

dft = dfUno.loc[dfUno['wc'].shift().eq('F4')]

But I would like to see it in a more generic way, may be extracting using iterrows() . Like, my code is only true for F4 . I want to extract the dataframe used for above F4/ F3 also.

Expected outcome for this:

 final  raw  act    wc             Start            Finish
   abc  xyz   30    M5  17-01-2022 06:00  14-07-2031 02:36
   abc  xyz   40    F4  17-01-2022 06:00  14-07-2031 02:36
   abc  pqr   40  M14S  17-01-2022 06:00  18-01-2026 17:21
   abc  pqr   50   M12  17-01-2022 06:00  18-01-2026 17:21
   abc  pqr   60  M14S  17-01-2022 06:00  18-01-2026 17:21
   abc  pqr   20    F3  17-01-2022 06:00  14-07-2031 02:36
   abc  pqr   40    F4  17-01-2022 06:00  14-07-2031 02:36

pls suggest something, how to do it.

Answer 1

You can first subset the frame to only have rows with wc equal to "F3" or "F4" . Then group by the raw column to see which index per group gives a maximum act . Then index the original frame with them:

>>> df.loc[df[df.wc.isin(["F3", "F4"])].groupby("raw", sort=False).act.idxmax()]

  final  raw  act  wc             Start            Finish
1   abc  xyz   40  F4  17-01-2022 06:00  14-07-2031 02:36
8   abc  pqr   40  F4  17-01-2022 06:00  14-07-2031 02:36

ie,

>>> df.wc.isin(["F3", "F4"]) 
0    False
1     True
2    False
3    False
4    False
5    False
6    False
7     True
8     True
9    False
Name: wc, dtype: bool

>>> subset = df[df.wc.isin(["F3", "F4"])]
>>> subset
  final  raw  act  wc             Start            Finish
1   abc  xyz   40  F4  17-01-2022 06:00  14-07-2031 02:36
7   abc  pqr   20  F3  17-01-2022 06:00  14-07-2031 02:36
8   abc  pqr   40  F4  17-01-2022 06:00  14-07-2031 02:36

>>> idxmax_act_per_raw = subset.groupby("raw", sort=False).act.idxmax()
>>> idxmax_act_per_raw 
raw
xyz    1
pqr    8
Name: act, dtype: int64

>>> df.loc[idxmax_act_per_raw]
  final  raw  act  wc             Start            Finish
1   abc  xyz   40  F4  17-01-2022 06:00  14-07-2031 02:36
8   abc  pqr   40  F4  17-01-2022 06:00  14-07-2031 02:36

(The sort=False of groupby assures raw isn't sorted whilst grouping, otherwise we'd get the row with "pqr" first at the end since "pqr" < "xyz" .)