Open jchester opened 2 months ago
N⁺/N⁻ have almost the same problem as C⁺/C⁻ had, which is that the ideal solution would be a field that is defined in terms of itself. They also share that if PostgreSQL had a reset when
on window functions, as Teradata does, the problem would be much easier to solve.
For now I am going to break out N⁺/N⁻ into their own issue so that I can move on to configurable decision intervals (H).
I think it would be best to hold off on this until #22 is done, because the way that sample averages are joined will change substantially.
individual_measurement_statistics_ewma
as it is now used in Cusum calculations as well.I won't do fast initial response in this one, but may reserve it for later. No intention of ever trying figure out how to do a V-mask procedure.