0

I am puzzled that you don't use ramp functions instead of step functions, then the estimation would be a purely linear regression/glm, and no need for gamma calculation. I assumed the optimisation in bayesian context would also be easier.

…**read more**, vote or answer

waiting for moderation