[Roadmap] 1.1.0 Roadmap

[hcho3]

@dmlc/xgboost-committer please add your items here by editing this post. Let's ensure that

• Each item has to be associated with a ticket
• Major design/refactoring are associated with a RFC before committing the code
• Blocking issue must be marked as blocking
• Breaking change must be marked as breaking

For other contributors who have no permission to edit the post, please comment here about what you think should be in 1.1.0.

• Add Survival analysis objective: Accelerated Failure Time (Add Accelerated Failure Time loss for survival analysis task #4763)
• Optimized ApplySplit and UpdatePredictCache functions on CPU (Optimized ApplySplit and UpdatePredictCache functions on CPU #5244). This concludes the 'hist' optimization work by @SmirnovEgorRu. Roadmap: CPU optimizations - 'hist' method #5104
• Refactored setup.py (Rewrite setup.py. #5271, Fix CMake + MSVC build with setuptools. #5280)
• Gradient based sampling for GPU Hist (Gradient based sampling for external memory mode on GPU #5093)
• Wide dataset quantile performance improvement (Wide dataset quantile performance improvement #5306)
• Thread safe in-place prediction (Thread safe, inplace prediction. #5389).
• JSON serialization of R XGBoost object ([R] Basic implementation for R JSON serialization. #5123)
• DMatrix refactoring (WIP) (Predict on Ellpack. #5327, Move prediction cache to Learner. #5220, Pass shared pointer instead of raw pointer to Learner. #5302, Remove update prediction cache from predictors. #5312, Remove SimpleCSRSource #5315, Refactor SparsePageSource, delete cache files after use #5321). RFC [RFC] Possible DMatrix refactor #4354, Roadmap [Roadmap] DMatrix refactoring #5143
• Pass feature count from JVM to native layer ([jvm-packages] Add JVM_CHECK_CALL. #5199, Add number of columns to native data iterator. #5202, [jvm-packages]add feature size for LabelPoint and DataBatch #5303)
• Deterministic GPU Histogram building (Model reproduciblility with histogram tree method. #5023).
• Use configured header in CMake (Ensure that configured dmlc/build_config.h is picked up by Rabit and XGBoost #5514)
• Parameter validation for scikit-learn and R (Enable parameter validation for skl. #5477, Enable parameter validation for R. #5569)

[hcho3]

Proposal: this time, let us not wait too long until the next release.

[hcho3]

@JohnZed @datametrician

[avinashbarnwal]

Please Add Survival analysis objective: Accelerated Failure Time #4763

[datametrician]

LGTM especially #5023. Thanks @hcho3!

[hcho3]

@avinashbarnwal See the first item in the list.

[tqchen]

@hcho3 would be great if we can look into apache/tvm#4953

[hcho3]

@tqchen I will look.

[RAMitchell]

@CodingCat can please you help us resolve the last JVM issue for the release?

[trivialfis]

Reverting the status for DMatrix refactoring.

[hcho3]

@trivialfis Is DMatrix refactor blocking 1.1.0 release?

[trivialfis]

@hcho3 I just want #5504 before 1.1, the PR fixes prediction on device dmatrix which is part of dmatrix refactoring. Also, the weighted sketching is not yet implemented for device dmatrix, but this is not blocking.

[trivialfis]

Also please let me give a deeper look into #5285 . Will continue profiling in coming days.