[SPARK-7770] Change GBT validationTol to be relative tolerance

• Type: Improvement
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Fix Version/s: 1.6.0
• Component/s: ML, MLlib
• Target Version/s: 1.6.0
• Assignee: Yanbo Liang
• Reporter: Joseph K. Bradley
• Shepherd: Joseph K. Bradley
• Created: 20/May/15 21:11
• Updated: 08/Oct/15 18:28
• Resolved: 08/Oct/15 18:28

Description

In spark.mllib, GBT validationTol uses absolute tolerance. Relative tolerance is arguably easier to set in a meaningful way. Questions:

• Should we change spark.mllib's validationTol meaning?
• Should we use relative tolerance in spark.ml's GBT (once we add validation support)?

I would vote for changing both to relative tolerance, where the tolerance is relative to the current loss on the training set.

Issue Links

• is related to: SPARK-5436 Validate GradientBoostedTrees during training
• is related to: SPARK-7132 Add fit with validation set to spark.ml GBT
• links to: [Github] Pull Request #8549 (yanboliang)