Keep maximum row in consecutive rows with specific value in column based on another column value-CodePudding

I have a df.

import pandas as pd 
df = pd.DataFrame({'id_c':[1] * 4   [2] * 3   [3] * 4,
        'Run':[7,8,5,4,3,2,1,2,3,4,5], 
      'Date_diff':[4,12,0,0,2,2,10,1,1,3,3]})

id_c   Run   Date_diff

1    7      4
1    8      12
1    5      0
1    4      0
2    3      2
2    2      2
2    1      10
3    2      1
3    3      1
3    4      3
3    5      3

For each unique value of id_c , if Date_diff equals to 0 , 1 , 2 for two consecutive rows , I want to keep the row with the maximum value in Run.

I tried :

df.groupby(['id_c' , 'Date_diff'])['Run'].idxmax()]

But it also selects maximum values for values of Date_diff different than 0 , 1 , 2.

The desired output would be :

id_c Run   Date_diff
1    7      4
1    8      12
1    5      0
2    3      2
2    1      10
3    3      1
3    4      3
3    5      3

Thanks!

CodePudding user response：

IIUC, compute a custom group and get the max index per group, then slice:

# get values not in 0/1/2
mask = ~df['Date_diff'].isin([0,1,2])
# group the consecutive 0/1/2 and get id of max Run
idx = df.groupby(['id_c', (mask|mask.shift()).cumsum()])['Run'].idxmax().values

# slice output with max ids
out = df.loc[idx]

output:

    id_c  Run  Date_diff
0      1    7          4
1      1    8         12
2      1    5          0
4      2    3          2
6      2    1         10
8      3    3          1
9      3    4          3
10     3    5          3