Group a timestamp-indexed series by hours/days

Question

I have a series indexed by timestamp

    date    value
    t1      x1
    t2      x2
    t3      x3
    ...

I would like to group the series by hours or days and apply a function group-wise that calculates the ratio

(value of the last timestamp in group - value of the first timestamp in group) / (last timestamp - first timestamp)

How can I do this?

possible duplicate of [Python Pandas: Group datetime column into hour and minute aggregations](http://stackoverflow.com/questions/16266019/python-pandas-group-datetime-column-into-hour-and-minute-aggregations) - you can see an example of how to solve this in the answer to the question — Jeff Tratner, Aug 10 '14 at 03:34
@JeffTratner it is not exact a duplicate, because I need also a way to grab the datetime start/end information in each group Thank you — Hello lad, Aug 10 '14 at 04:20
the difficult part of the problem is to do that grouping, the datetime info in the start and end you can g — Jeff Tratner, Aug 10 '14 at 04:45

score 0 · Accepted Answer · answered Aug 10 '14 at 08:39

0

(last timestamp - first timestamp)

I'm assuming you mean value of timestamp?

It's possible I've misinterpreted the problem, but this might give you an idea.

df.resample('1H', how=lambda x: (x[-1] - x[0]) / (df[-1] - df[0]))

answered Aug 10 '14 at 08:39

Easypeasy

1 Answers1