I suspect many people working on timeseries data have already come across this issue, and pandas doesn\'t seem to provide a straightforward solution (yet!):
Suppose:
