For the extended MSTATS some of the standard (COLLAPSE etc.) estimators
don't strictly apply. These are the ones that use the collapse-axis
co-ordinates, such as "Iwc", "Comax", and "Integ". Now in my first
draft I excluded these. However, they could be retained by generating a
pseudo-axis of indices based upon the order of the supplied NDFs. We
could even use some other metadata (like a time), but those would have
to be validated as monotonic or sorted. This possible feature would be
useful for a time series of frames.
So should I continue with Plan A to exclude the estimators tied to
co-ordinate, leaving that until there's a specific call for it?
Malcolm
|