Documentation: clarification of the estimate argument for each metric

[RandallJEllis]

Problem
I spent a day trying to understand the cumulative_dynamic_auc function because I was consistently getting values in the 0.15-0.3 range. After experimenting, I realized that for the estimate argument, I was using the output of predict_survival_function (in a RandomSurvivalForest model) instead of predict, and switching this fixed the issue and gave me reasonable AUCs.

Proposed solution
The documentation for cumulative_dynamic_auc is not very descriptive on what kind of outputted predictions (e.g., probabilities, estimated times-to-event) the estimate argument should be, and I think this could be clearer for all metrics.

[sebp]

Thanks for reporting this issue. The current description for estimate is:

Estimated risk of experiencing an event of test data. If estimate is a 1-d array, the same risk score across all time points is used. If estimate is a 2-d array, the risk scores in the j-th column are used to evaluate the j-th time point.

It does mention that estimate should be a risk score.

Can you please elaborate which parts are not clear to you? Is it that the description isn't referencing predict?