We're updating the issue view to help you get more done. 

Custom metric not evaluated in internal mode of Sparkling Water

Description

If the attached 'internal_custom_metric.py' file is run on EMR+SW (internal mode only) it prints:

1 2 None 0.0

When it should print something like:

1 2 WeightedFalseNegativeLoss 0.42381109064735906

Issue is apparent with python 3.6 and 2.7.

Old Description
Using raw code from this notebook: https://github.com/h2oai/h2o-tutorials/blob/master/tutorials/custom_metric_func/CustomMetricFuncClassification.ipynb

1 gbm_v2.model_performance()

doesn't print the custom metric.

Code below has all 0s for the custom metric score:

1 sorted([[h2o.get_model(x).model_performance(xval = True).custom_metric_value(), x] for x in gbm_grid.model_ids])

Environment

None

Status

Assignee

Jakub Hava

Reporter

Joseph Granados

Labels

None

Release Priority

None

CustomerVisible

Yes

testcase 1

None

testcase 2

None

testcase 3

None

h2ostream link

None

Affected Spark version

None

AffectedContact

None

AffectedCustomers

None

AffectedPilots

None

AffectedOpenSource

None

Support Assessment

None

Customer Request Type

None

Support ticket URL

None

End date

None

Baseline start date

None

Baseline end date

None

Task progress

None

Task mode

None

Fix versions

Priority

Major