Good question. We should probably provide more description/documentation.

There is a paper that describes the residential stochastic occupancy schedules here:

• Stochastic simulation of occupant-driven energy use in a bottom-up residential building stock model

(The implementation of this methodology -- used by BEopt, ResStock, URBANopt, etc. -- can be found in an OpenStudio measure in the OpenStudio-HPXML repository.)