Page MenuHomePhabricator

Consistent use of attribute indicating whether small metric values lead to a better rank
Closed, ResolvedPublic


In the package the two variables smallBetter and largeBetter are used. Should this be unified as this is error-prone? Developers always have to take care to invert the variable respectively.

Event Timeline

I would propose to go with largeBetter, as this is true for the most common metric values (e.g. DSC and Accuracy).

yes this has historical reasons (largeBetter had been "inverseOrder" in the beginning which was horrible;-))
smallBetter would have the advantage that the user interface would not change, modifying as.challenge with largeBetter would mean that existing scripts lead to wrong results which is problematic.
could someone of you please take care of this one?

wiesenfa lowered the priority of this task from Normal to Low.Aug 31 2020, 2:45 PM