[SPARK-19357] Parallel Model Evaluation for ML Tuning: Scala

• Type: Sub-task
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 2.3.0
• Component/s: ML
• Assignee: Bryan Cutler
• Reporter: Bryan Cutler
• Created: 25/Jan/17 01:10
• Updated: 01/Jan/18 01:00
• Resolved: 06/Sep/17 12:12

Description

This is a first step of the parent task of Optimizations for ML Pipeline Tuning to perform model evaluation in parallel. A simple approach is to naively evaluate with a possible parameter to control the level of parallelism. There are some concerns with this:

• excessive caching of datasets
• what to set as the default value for level of parallelism. 1 will evaluate all models in serial, as is done currently. Higher values could lead to excessive caching.

Issue Links

• is related to: SPARK-21911 Parallel Model Evaluation for ML Tuning: PySpark
• is related to: SPARK-22126 Fix model-specific optimization support for ML tuning
• relates to: SPARK-21027 Parallel One vs. Rest Classifier
• relates to: SPARK-19979 [MLLIB] Multiple Estimators/Pipelines In CrossValidator
• links to: [Github] Pull Request #16774 (BryanCutler)