Pandas 数据帧中任意两连续行之间差异的平均值

Question

I have a dataframe

name   date         quantity
'A'    2016-12-02   20
'A'    2016-12-04   5
'A'    2016-11-30   10
'B'    2016-11-30   10
...

What I want to do is calculate, for any pair of consecutive dates (consecutive as in chronological) for a name, the difference in the quantity, and the average these counts for a name.

Dates are indeed not necessarily presented in a chronological ordering.

Specifically, for name A I'd want to compute +10 (difference 2nd Dec - 30 Nov) and -15 (difference 4th Dec -2nd Nov) and then average those, to get a final result of -2.5 for this name.

Ideas?

Answer 1

You can use groupby and apply diff with mean :

print (df.groupby('name')['quantity'].apply(lambda x: x.diff().mean()).reset_index())
  name  quantity
0  'A'      -2.5
1  'B'       NaN

EDIT: You can add sort_values by column date

print (df.sort_values('date')
         .groupby('name')['quantity']
         .apply(lambda x: x.diff().mean())
         .reset_index())
  name  quantity
0  'A'      -2.5
1  'B'       NaN