Closed mllg closed 1 year ago
@RaphaelS1 does this solve #816 ?
Looks like it should. Happy for @sebffischer to test and decide if it makes sense (saw his ideas in the other issues) and approve if so
So @mllg s solution is much simpler than my suggestion. It also solves the bug from the other issue:
library(mlr3proba)
#> Loading required package: mlr3
library(mlr3extralearners)
l = lrn("surv.ranger")
t = tsk("rats")
l$train(t)
p = l$predict(t)
m = msr("oob_error")
p$score(m, learner = l)
#> oob_error
#> 0.2516677
l$oob_error()
#> [1] 0.2516677
m$predict_type
#> [1] NA
p$predict_types
#> [1] "crank" "distr"
Created on 2023-01-24 by the reprex package (v2.0.1)
Addressing #816
Change
predict_type
of generic measures toNA
to avoid an unnecessary check which renders extensions difficult.Also renamed
MeasureDebug
to be more specific.