The "fp" command for fractional polynomial regression allows, for a single explanatory variable "x", in case of a maximum possible degree (option "dimension") above 1 (and the default is 2), repeated powers (by multiplying every repetition by log(x)). Can repeated-power combinations be excluded? So, for example, in case powers 0.5, 1 and 1.5 are selected, is it possible to ask the algorithm to just explore 3 second degree combinations (sqrt(x) and x; sqrt(x) and x2; x and x2) instead of 6 (thus, ignoring x and x*log(x), x2 and x2*log(x), sqrt(x) and sqrt(x)*log(x))?
-
Login or Register
- Log in with
