Probability thresholding should be explained in a better way in help about mlr HOT 3 CLOSED

I suppose you are talking about setThreshold()?
Is this already existant explanaiton in the tutorial sufficiant?
I will improve the R-Help

from mlr.

Hmm. It should be mentioned in that section, yes. But maybe we need a different, separate section that goes into more details.

a) First, regarding that section:

• It should be explained how for predict.type = "prob" the class labels are generated. These come now out of the predicted probabilities and the default threshold.
• Thresholding works also for more than 2 classes!

b)
Here are all function that deal deal with predicted probabilities:

• setThreshold, also can be used on a prediction after resampling!
• getProbabilities
• tuneThreshold
• asROCRPrediction, but we have a section on this. Could be referenced if we write this down in one additional part.

from mlr.

I will do the following:
Prediction: short mention of setThreshold / getProbabliies
ROC Analysis: mention the rest.

from mlr.