Github user dbtsai commented on a diff in the pull request: https://github.com/apache/spark/pull/14834#discussion_r78419383 --- Diff: mllib/src/main/scala/org/apache/spark/ml/classification/LogisticRegression.scala --- @@ -311,8 +350,28 @@ class LogisticRegression @Since("1.2.0") ( val histogram = labelSummarizer.histogram val numInvalid = labelSummarizer.countInvalid - val numClasses = histogram.length val numFeatures = summarizer.mean.size + val numFeaturesPlusIntercept = if (getFitIntercept) numFeatures + 1 else numFeatures + + val numClasses = MetadataUtils.getNumClasses(dataset.schema($(labelCol))) match { + case Some(n: Int) => + require(n >= histogram.length, s"Specified number of classes $n was " + + s"less than the number of unique labels ${histogram.length}.") + n + case None => histogram.length + } + + val isBinaryClassification = numClasses == 1 || numClasses == 2 + val isMultinomial = $(family) match { + case "binomial" => + require(isBinaryClassification, s"Binomial family only supports 1 or 2 " + + s"outcome classes but found $numClasses.") + false + case "multinomial" => true + case "auto" => !isBinaryClassification + case other => throw new IllegalArgumentException(s"Unsupported family: $other") + } --- End diff -- Oh, I original thought it was a typo. `isBinaryClassification` is not used in other place, and I think it's kind of leading people to interpret as `isBinomial`. Maybe we should just remove it.
--- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. --- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org