SVM: binomial model - AUC curves are missing

Description

SVM does not contain AUC for train/validation

when
```
buildModel 'svm', {"model_id":"svm-mm","training_frame":"train","response_column":"Score","validation_frame":"validation","nfolds":0,"ignored_columns":[],"ignore_const_cols":true,"add_intercept":true,"step_size":"10","reg_param":0.01,"convergence_tol":0.001,"mini_batch_fraction":".2","threshold":0,"updater":"Simple","gradient":"LeastSquares”}```

not curves are generated

Activity

Show:
Mateusz Dymczyk
October 19, 2016, 1:49 PM

Pushed the fix into h2o-flow

Michal Malohlava
October 5, 2016, 5:24 PM

Yup, i was doing manual prediction as a workaround. I did not know that it is hardcoded in Flow. The good person is Prithvi, or you can make PR to https://github.com/h2oai/h2o-flow

Mateusz Dymczyk
October 5, 2016, 2:57 AM

I would need to talk about this with someone who's maintaining FlowUI - those curves show when you make predictions using SVM models. They don't appear right after training because it's hardcoded in one of the FlowUI coffeescript files to only appear (the ROC curve) for ~2 H2O models.

Fixed

Assignee

Mateusz Dymczyk

Reporter

Michal Malohlava

Labels

None

CustomerVisible

No

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

ReleaseNotesHidden

None

Fix versions

Affects versions

Priority

Major