H2O Binomial scorer

KCH Workflows
1

Dear all,

I am using ‘H2O binomial scorer’ node for evaluating a Random forest model. I want to know details about the various statistical parameters provided by the second output port of this node some of which are not clear to me (e.g. log loss, Max F1, etc.).
Where can I find more information for these parameters?

Thanks,
San

2

Hi @sundriyals -

As you’ve probably noticed, this particular node outputs a lot of different statistics. :slight_smile: Unfortunately, we don’t have a comprehensive list available in the description, so your best bet here is probably Google. Fortunately, these are all pretty standard metrics, so info should be easy to come by.

Was there one statistic in particular you were interested in using?

3

Thanks @ScottF

Not looking for any specific parameter.
I just wanted more information about the ones calculated by this node. For e.g. their significance and calculation

4

Hello,

I have some questions regarding this topic. Does ‘MCC’ stand for Matthews correlation coefficient? And what are ‘Max F1 (AUC)’ and ‘Pr (AUC)’ respectively?

Thank you.

5

This page in the H2O docs will hopefully shed some additional light on the various metrics: Scorers — Using Driverless AI 1.9.2.1 documentation