[SPARK-9478] Add sample weights to Random Forest

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.4.1
• Fix Version/s: 3.0.0
• Component/s: ML
• Assignee: Ruifeng Zheng
• Reporter: Patrick Crenshaw
• Created: 30/Jul/15 15:07
• Updated: 15/Jan/20 03:36
• Resolved: 14/Jan/20 14:27

Description

Currently, this implementation of random forest does not support sample (instance) weights. Weights are important when there is imbalanced training data or the evaluation metric of a classifier is imbalanced (e.g. true positive rate at some false positive threshold). Sample weights generalize class weights, so this could be used to add class weights later on.

Issue Links

• duplicates: SPARK-19591 Add sample weights to decision trees
• is blocked by: SPARK-19591 Add sample weights to decision trees
• is blocked by: SPARK-14599 BaggedPoint should support weighted instances.
• Is contained by: SPARK-14046 RandomForest improvement umbrella
• relates to: SPARK-9610 Class and instance weighting for ML
• links to: [Github] Pull Request #9008 (rotationsymmetry)
• links to: [Github] Pull Request #13851 (n-triple-a)
• links to: [Github] Pull Request #16722 (sethah)
• links to: GitHub Pull Request #27097