machine learning - Limiting prediction range for Weka -
for research using weka predict alpha values different uses. legal range of alpha real number between 0 , 1 inclusive. performing well, of predictions greater 1. want keep classifier numerical since real number, want limit range of prediction between 0 , 1. ideas on how this?
i think @lars-kotthoff raises interesting points. provide suggestions different perspective, ignoring classification problems:
once have set of values within range [0, inf), can try normalised them using function such logit or min-max, among others.
Comments
Post a Comment