[SPARK-31681] Python multiclass logistic regression evaluate should return LogisticRegressionSummary

• Type: Bug
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: 3.0.0
• Fix Version/s: 3.0.0
• Component/s: ML, PySpark
• Labels: release-notes

Docs Text

In Spark 3.0, a multiclass logistic regression in Pyspark will now (correctly) return LogisticRegressionSummary, not the subclass BinaryLogisticRegressionSummary. The additional methods exposed by BinaryLogisticRegressionSummary would not work in this case anyway.

• Assignee: Huaxin Gao
• Reporter: Huaxin Gao
• Created: 11/May/20 22:09
• Updated: 14/May/20 15:55
• Resolved: 14/May/20 15:55

Description

    def evaluate(self, dataset):
        ......
        java_blr_summary = self._call_java("evaluate", dataset)
        return BinaryLogisticRegressionSummary(java_blr_summary)

We should return LogisticRegressionSummary instead of BinaryLogisticRegressionSummary for multiclass LogisticRegression

Issue Links

• links to: [Github] Pull Request #28503 (huaxingao)
• links to: [Github] Pull Request #28503 (huaxingao)