[SPARK-6004] Pick the best model when training GradientBoostedTrees with validation

• Type: Improvement
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: 1.4.0
• Fix Version/s: 1.4.0
• Component/s: MLlib
• Target Version/s: 1.4.0
• Assignee: L. C. Hsieh
• Reporter: L. C. Hsieh
• Created: 25/Feb/15 10:41
• Updated: 26/Feb/15 19:04
• Resolved: 26/Feb/15 19:04

Description

Since the validation error does not change monotonically, in practice, it should be proper to pick the best model when training GradientBoostedTrees with validation instead of stopping it early.

Issue Links

• links to: [Github] Pull Request #4763 (viirya)