Aggregate time series in python

Question

How do we aggregate the time series by hour or minutely granularity? If I have a time series like the following then I want the values to be aggregated by hour. Does pandas support it or is there a nifty way to do it in python?

timestamp, value
2012-04-30T22:25:31+00:00, 1
2012-04-30T22:25:43+00:00, 1
2012-04-30T22:29:04+00:00, 2
2012-04-30T22:35:09+00:00, 4
2012-04-30T22:39:28+00:00, 1
2012-04-30T22:47:54+00:00, 8
2012-04-30T22:50:49+00:00, 9
2012-04-30T22:51:57+00:00, 1
2012-04-30T22:54:50+00:00, 1
2012-04-30T22:57:22+00:00, 0
2012-04-30T22:58:38+00:00, 7
2012-04-30T23:05:21+00:00, 1
2012-04-30T23:08:56+00:00, 1

I also tried to make sure I have the correct data types in my data frame by calling:

  print data_frame.dtypes

and I get the following as out put

ts     datetime64[ns]
val             int64

When I call group by on the data frame

grouped = data_frame.groupby(lambda x: x.minute)

I get the following error:

grouped = data_frame.groupby(lambda x: x.minute)
AttributeError: 'int' object has no attribute 'minute'

Solution

http://pandas.pydata.org/pandas-docs/dev/generated/pandas.DataFrame.resample.html DataFrame.resample method. You can specify here way of aggregation, in your case sum.

data_frame.resample("1Min", how="sum")

http://pandas.pydata.org/pandas-docs/dev/timeseries.html#up-and-downsampling