i have a binary classification problem so i am using xgbclassifier. i set the objective function to binary:logistic. Therefore when i employ: predict.proba(x_test) are these probabilites or scores?? would you recommond calibration
Predict.proba() output calibration
How well an estimator generates its probabilities is something you can explore (and test), see https://scikit-learn.org/stable/modules/calibration.html for a discussion.
1 Like