Pandas: Difference to previous value

Question

Given a Pandas Data Frame that looks like this

GROUP   VALUE    MASK
  1        5     false
  2       10     false
  2       20     false
  1        7     tr         


        

Answer 1

use groupby, diff, 'MASK'

pd.concat([df.VALUE, df.groupby('GROUP').VALUE.diff()],
          axis=1, keys=['VALUE', 'DIFF'])[df.MASK]