[SPARK-12664] Expose probability, rawPrediction in MultilayerPerceptronClassificationModel

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 2.3.0
• Component/s: ML
• Assignee: Weichen Xu
• Reporter: Robert Dodier
• Shepherd: Joseph K. Bradley
• Created: 06/Jan/16 01:24
• Updated: 29/Aug/17 04:11
• Resolved: 23/Aug/17 04:16

Description

In org.apache.spark.ml.classification.MultilayerPerceptronClassificationModel, there isn't any way to get raw prediction scores; only an integer output (from 0 to #classes - 1) is available via the `predict` method.

`mplModel.predict` is called within the class to get the raw score, but `mlpModel` is private so that isn't available to outside callers.

The raw score is useful when the user wants to interpret the classifier output as a probability.

Issue Links

• is depended upon by: SPARK-21856 Update Python API for MultilayerPerceptronClassifierModel
• is related to: SPARK-9471 Multilayer perceptron classifier
• is related to: SPARK-21802 Make sparkR MLP summary() expose probability column
• is related to: SPARK-21729 Generic test for ProbabilisticClassifier to ensure consistent output columns
• links to: [Github] Pull Request #17373 (WeichenXu123)