Pandas groupby with each group treated as a unique group-CodePudding

Please assist. How do I get the cumsum of a pandas groupby, but my data is boolean 0 and 1. I want to treat each group of 0s or 1s as unique values, and the count to reset when new values are met.

I currently have this which sums up all 1s and 0's

df['grp'] = df.groupby("dir")["dir"].cumsum())

My desired output

df  = pd.DataFrame({"dir":[1,1,1,1,0,0,0,1,1,1,1,0,0,0],
                  "grp": [1,2,3,4,1,2,3,1,2,3,4,1,2,3,]})

CodePudding user response：

Use:

In [1495]: df['grp'] = df.groupby((df['dir'] != df['dir'].shift(1)).cumsum()).cumcount() 1

In [1496]: df
Out[1496]: 
    dir  grp
0     1    1
1     1    2
2     1    3
3     1    4
4     0    1
5     0    2
6     0    3
7     1    1
8     1    2
9     1    3
10    1    4
11    0    1
12    0    2
13    0    3