Closed russelljjarvis closed 7 years ago
@russelljjarvis The sort_key
should always be a value (or array of values) between 0 and 1, with 1 being good (e.g. Z=0) and 0 being bad (e.g. Z=very positive or very negative). You shouldn't have to do any transformation of it, except perhaps multiplying by -1 if you are trying to minimize the objective function.
The transformation of e.g. Z to the sort_key
is done in sciunit.scores
, in a different way for each score type. If you are having trouble with any of these we can consider changing it, or you can override it in your own custom Score
type.
Hi @rgerkin I just wanted to double check that goodness in the context of
score.sort_key.values
, is distance from zero in both positive and negative directions? If so when my optimizer encounters a score, as its job is to minimize badness, I divide 1 by the absolute value of the score, and multiply by -1.0. In this way I intend to convert the goodness to absence of badness.