BME analysis was a good start. The ability to move boundaries and find the optimal boundary in terms of difference of flux/entropy/etc would be better.
The simplest case is just BE - single fold performance structure maybe? think of a better name.
if p(x,y) takes the proportion of a performance between x and y (x and y in [0,1])
So we need to look for the w in [0,1] such that the following expression is maximised:
abs(flux(p(0,w)) - flux(p(w,1)))
Maybe in practice, we could just sample for 100-1000 values of w in each performance and plot this for all performances to get a feel for the possible distributions.
BME analysis was a good start. The ability to move boundaries and find the optimal boundary in terms of difference of flux/entropy/etc would be better.
The simplest case is just BE - single fold performance structure maybe? think of a better name.
if p(x,y) takes the proportion of a performance between x and y (x and y in [0,1])
So we need to look for the w in [0,1] such that the following expression is maximised:
abs(flux(p(0,w)) - flux(p(w,1)))
Maybe in practice, we could just sample for 100-1000 values of w in each performance and plot this for all performances to get a feel for the possible distributions.