Description of estimate parameter in brier_score is unclear about scikit-survival HOT 4 CLOSED

Fixed in #424

from scikit-survival.

Thanks for helping to improve scikit-survival.

Could your please elaborate which part you consider confusing and why, please?

from scikit-survival.

Thanks for getting back!
The brier score functions take as estimate input the probability for the negative class while the other metrics take the probability (or score) for the positive class (=event). The first line describing the estimate, however, is the same: estimate: Estimated risk of experiencing an event for test data. I would then believe, that a high estimate should - consistently with AUC_C,D - correspond to the positive class. However, reading the second sentence then clarifies that it is actually the other way around: The i-th column must contain the estimated probability of remaining event-free up to the i-th time point.

I thought the two sentences are unnecessarily ambiguous and could lead to some confusion.

from scikit-survival.

Duplicate with #414 i guess

from scikit-survival.