Ah! We could, but StatsD provides support for more complex metrics like aggregated sums over time. Not something that lends itself easily to a discrete datapoin

Ah! We could, but StatsD provides support for more complex metrics like aggregated sums over time. Not something that lends itself easily to a discrete datapoint. It is. But we use multiprocessing, which is basically the same API. Still, you can't beat the awesome Python stats libraries: Numpy, SciPy, Statsmodels, Pandas..