Calculate difference of adjacent rows (decimal numbers) in a data frame for each group defined in a different column

Question

I have a data frame with three columns of interest, 'time', 'peak' and 'cycle'. I want to calculate the time elapsed between each row for a given cycle.

   time  peak  cycle
0     1     1      1
1     2     0      1
2   3.5     0      1
3   3.8     1      2
4     5     0      2
5   6.2     0      2
6     7     0      2

I want to add a fourth column, so the data frame would look like this when complete:

   time  peak  cycle  time_elapsed
0     1     1      1             0
1     2     0      1             1
2   3.5     0      1           1.5
3   3.8     1      2             0
4     5     0      2           1.2
5   6.2     0      2           1.2
6     7     0      2           0.8

The cycle number is calculated based on the peak information, so I don't think I need to refer to both columns.

data['time_elapsed'] = data['time'] - data['time'].shift()

Applying the above code I get:

   time  peak  cycle  time_elapsed
0     1     1      1             0
1     2     0      1             1
2   3.5     0      1           1.5
3   3.8     1      2           0.3
4     5     0      2           1.2
5   6.2     0      2           1.2
6     7     0      2           0.8

Is there a way to "reset" the calculation every time the value in 'peak' is 1?Any tips or advice would be appreciated!

Answer 1

Subtract first value per groups converted in Series by GroupBy.transform with GroupBy.first:

df['time_elapsed'] = df['time'].sub(df.groupby('cycle')['time'].transform('first'))
print (df)
   time  peak  cycle  time_elapsed
0     1     1      1             0
1     2     0      1             1
2     3     0      1             2
3     4     1      2             0
4     5     0      2             1
5     6     0      2             2
6     7     0      2             3

For adding reset add new Series with Series.cumsum - if values are only 1 or 0 in peak column:

s = df['peak'].cumsum()
df['time_elapsed'] = df['time'].sub(df.groupby(['cycle', s])['time'].transform('first'))