Suggestion: convert to time series

[tschundler]

Since the data in graphite is time series data, IMO it makes sense to have it in a DatetimeIndex.

I used this, which works, and may be worth doing by default:

        data.index = pd.DatetimeIndex(data.index.map(datetime.fromtimestamp))

Something like this should be more efficient, but gave me timezone problems:

        data.index = pd.DatetimeIndex(data.index.values.astype('datetime64[s]'))

Anyways, just a consideration. It doesn't really get in the way of use.