[ https://issues.apache.org/jira/browse/IGNITE-9421?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16596243#comment-16596243
]
Oleg Ignatenko commented on IGNITE-9421:
----------------------------------------
per preliminary discussion with [~zaleslaw] this is most likely bug in the model, it should
round results. This looks plausible because very similar example {{LogRegressionMultiClassClassificationExample}}
using a similar model runs fine, meaning that other model is implemented correctly.
> ML Examples: LogisticRegressionSGDTrainerExample example result not correct
> ---------------------------------------------------------------------------
>
> Key: IGNITE-9421
> URL: https://issues.apache.org/jira/browse/IGNITE-9421
> Project: Ignite
> Issue Type: Bug
> Components: ml
> Affects Versions: 2.6
> Reporter: Stepan Pilschikov
> Assignee: Aleksey Zinoviev
> Priority: Major
> Fix For: 2.7
>
>
> Running org.apache.ignite.examples.ml.regression.logistic.binary.LogisticRegressionSGDTrainerExample
example
> Output:
> {code}
> >>> Absolute amount of errors 100
> >>> Accuracy 0.0
> >>> Confusion matrix is [[50, 50], [0, 0]]
> >>> ---------------------------------
> >>> Logistic regression model over partitioned dataset usage example completed.
> {code}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
|