This is in reference to #97. The brier score according to https://en.wikipedia.org/wiki/Brier_score indicates that scores range from 0 to 1 and the equation is a squared-value so this makes sense.
When interpreting the brier scores between models, how does one interpret negative range of scores?
I am running the ml_predictor.train function for LGBMClassifier, RandomForestClassifier, and CatBoostClassifier methods which all return a negative score. CatBoost then is chosen as the "best" model which is used to train the model then at the end returns a CV score which is also negative.
I am new to auto_ml and wanted to make sure I am interpreting the results according to the intended way. Thanks for your help!
This is in reference to #97. The brier score according to https://en.wikipedia.org/wiki/Brier_score indicates that scores range from 0 to 1 and the equation is a squared-value so this makes sense.
When interpreting the brier scores between models, how does one interpret negative range of scores?
I am running the ml_predictor.train function for LGBMClassifier, RandomForestClassifier, and CatBoostClassifier methods which all return a negative score. CatBoost then is chosen as the "best" model which is used to train the model then at the end returns a CV score which is also negative.
I am new to auto_ml and wanted to make sure I am interpreting the results according to the intended way. Thanks for your help!
-Carter