[SPARK-14829] Deprecate GLM APIs using SGD

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 2.0.0
• Component/s: MLlib
• Target Version/s: 2.0.0
• Assignee: Ruifeng Zheng
• Reporter: Joseph K. Bradley
• Shepherd: Joseph K. Bradley
• Created: 21/Apr/16 23:53
• Updated: 20/May/16 22:55
• Resolved: 29/Apr/16 05:44

Description

I don't know how many times I have heard someone run into issues with LinearRegression or LogisticRegression, only to find that it is because they are using the SGD implementations in spark.mllib. We should deprecate these SGD APIs in 2.0 to encourage users to use LBFGS and the spark.ml implementations, which are significantly better.

Issue Links

• relates to: SPARK-15457 Eliminate MLlib 2.0 build warnings from deprecations
• links to: [Github] Pull Request #12596 (zhengruifeng)