[
https://issues.apache.org/jira/browse/SPARK-9312?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14641303#comment-14641303
]
Joseph K. Bradley commented on SPARK-9312:
------------------------------------------
A few comments:
* Logistic regression provides predictive probabilities under the spark.ml
package. Will that suffice for your use case?
* For OneVsRest, I do not see a good way to predict probabilities. We could
output confidences (rawPrediction column, implemented using Classifier
abstraction), but it is not principled to interpret them as probabilities.
Could you please update your PR to do that instead?
* In general, it's good to separate issues, rather than combining them under
the same JIRA/PR.
Could you please separate these issues and refactor OneVsRest? Thanks!
> The multiclass logistic regression and the OneVsRest model do not provide
> predictive probability (confidence factor) along with the prediction
> ----------------------------------------------------------------------------------------------------------------------------------------------
>
> Key: SPARK-9312
> URL: https://issues.apache.org/jira/browse/SPARK-9312
> Project: Spark
> Issue Type: Improvement
> Components: ML, MLlib
> Affects Versions: 1.4.0, 1.4.1
> Reporter: Badari Madhav
> Labels: features
> Original Estimate: 72h
> Remaining Estimate: 72h
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
