Cuml Versions Save

🐛 Bug Fixes

• Skipping some hdbscan tests when cuda version is <= 11.2. (#4916) @cjnolet
• Fix HDBSCAN python namespace (#4895) @cjnolet
• Cupy 11 fixes (#4889) @dantegd
• Fix small fp precision failure in linear regression doctest test (#4884) @lowener
• Remove unused cuDF imports (#4873) @beckernick
• Update for thrust 1.17 and fixes to accommodate for cuDF Buffer refactor (#4871) @dantegd
• Use rapids-cmake 22.10 best practice for RAPIDS.cmake location (#4862) @robertmaynard
• Patch for nightly test&bench (#4840) @viclafargue
• Fixed Large memory requirements for SimpleImputer strategy median #4794 (#4817) @erikrene
• Transforms RandomForest estimators non-consecutive labels to consecutive labels where appropriate (#4780) @VamsiTallam95

📖 Documentation

• Document that minimum required CMake version is now 3.23.1 (#4899) @robertmaynard
• Update KMeans notebook for clarity (#4886) @beckernick

🚀 New Features

• Allow cupy 11 (#4880) @galipremsagar
• Add sample_weight to Coordinate Descent solver (Lasso and ElasticNet) (#4867) @lowener
• Import treelite models into FIL in a different precision (#4839) @canonizer
• #4783 Added nan_euclidean distance metric to pairwise_distances (#4797) @Sreekiran096
• PowerTransformer, QuantileTransformer and KernelCenterer (#4755) @viclafargue
• Add "median" to TargetEncoder (#4722) @daxiongshu
• New Feature StratifiedKFold (#3109) @daxiongshu

🛠️ Improvements

• Updating python to use pylibraft (#4887) @cjnolet
• Upgrade Treelite to 3.0.0 (#4885) @hcho3
• Statically link all CUDA toolkit libraries (#4881) @trxcllnt
• approximate_predict function for HDBSCAN (#4872) @tarang-jain
• Pin dask and distributed for release (#4859) @galipremsagar
• Remove Raft deprecated headers (#4858) @lowener
• Fix forward-merge conflicts (#4857) @ajschmidt8
• Update the NVTX bench helper for the new nsys utility (#4826) @viclafargue
• All points membership vector for HDBSCAN (#4800) @tarang-jain
• TSNE and UMAP allow several distance types (#4779) @tarang-jain
• Convert fp32 datasets to fp64 in ARIMA and AutoARIMA + update notebook to avoid deprecation warnings with positional parameters (#4195) @Nyrio

🚨 Breaking Changes

• Update Python build to scikit-build (#4818) @dantegd
• Bump xgboost to 1.6.0 from 1.5.2 (#4777) @galipremsagar

🐛 Bug Fixes

• Revert "Allow CuPy 11" (#4847) @galipremsagar
• Fix RAFT_NVTX option not set (#4825) @achirkin
• Fix KNN error message. (#4782) @trivialfis
• Update raft pinnings in dev yml files (#4778) @galipremsagar
• Bump xgboost to 1.6.0 from 1.5.2 (#4777) @galipremsagar
• Fixes exception when using predict_proba on fitted Pipeline object with a ColumnTransformer step (#4774) @VamsiTallam95
• Regression errors failing with mixed data type combinations (#4770) @shaswat-indian

📖 Documentation

• Use common code in python docs and defer js loading (#4852) @galipremsagar
• Centralize common css & js code in docs (#4844) @galipremsagar
• Add ComplementNB to the documentation (#4805) @lowener
• Fix forward-merge branch-22.06 to branch-22.08 (#4789) @divyegala

🚀 New Features

• Update Python build to scikit-build (#4818) @dantegd
• Vectorizers to accept Pandas Series as input (#4811) @shaswat-indian
• Cython wrapper for v-measure (#4785) @shaswat-indian

🛠️ Improvements

• Pin dask & distributed for release (#4850) @galipremsagar
• Allow CuPy 11 (#4837) @jakirkham
• Remove duplicate adj_to_csr implementation (#4829) @ahendriksen
• Update conda environment files to UCX 1.13.0 (#4813) @pentschev
• Update conda recipes to UCX 1.13.0 (#4809) @pentschev
• Fix #3414: remove naive versions dbscan algorithms (#4804) @ahendriksen
• Accelerate adjacency matrix to CSR conversion for DBSCAN (#4803) @ahendriksen
• Pin max version of cuda-python to 11.7.0 (#4793) @Ethyling
• Allow cosine distance metric in dbscan (#4776) @tarang-jain
• Unpin dask & distributed for development (#4771) @galipremsagar
• Clean up Thrust includes. (#4675) @bdice
• Improvements in feature sampling (#4278) @vinaydes

v22.06.01

🐛 Bug Fixes

• Fix sg benchmark build. (#4766) @trivialfis
• Resolve KRR hypothesis test failure (#4761) @RAMitchell
• Fix KBinsDiscretizer bin_edges_ (#4735) @viclafargue
• FIX Accept small floats in RandomForest (#4717) @thomasjpfan
• Remove import of scalar_broadcast_to from stemmer (#4706) @viclafargue
• Replace 22.04.x with 22.06.x in yaml files (#4692) @daxiongshu
• Replace cudf.logical_not with ~ (#4669) @canonizer

📖 Documentation

• Fix docs builds (#4733) @ajschmidt8
• Change "principals" to "principles" (#4695) @cakiki
• Update pydoc and promote ColumnTransformer out of experimental (#4509) @viclafargue

🚀 New Features

• float64 support in FIL functions (#4655) @canonizer
• float64 support in FIL core (#4646) @canonizer
• Allow "LabelEncoder" to accept cupy and numpy arrays as input. (#4620) @daxiongshu
• MNMG Logistic Regression (dask-glm wrapper) (#3512) @daxiongshu

🛠️ Improvements

• Pin dask & distributed for release (#4758) @galipremsagar
• Simplicial set functions (#4756) @viclafargue
• Upgrade Treelite to 2.4.0 (#4752) @hcho3
• Simplify recipes (#4749) @Ethyling
• Inference for float64 random forests using FIL (#4739) @canonizer
• MNT Removes unused optim_batch_size from UMAP's docstring (#4732) @thomasjpfan
• Require UCX 1.12.1+ (#4720) @jakirkham
• Allow enabling raft NVTX markers when raft is installed (#4718) @achirkin
• Fix identifier collision (#4716) @viclafargue
• Use raft::span in TreeExplainer (#4714) @hcho3
• Expose simplicial set functions (#4711) @viclafargue
• Refactor tests in cuml (#4703) @galipremsagar
• Use conda to build python packages during GPU tests (#4702) @Ethyling
• Update pinning to allow newer CMake versions. (#4698) @vyasr
• TreeExplainer extensions (#4697) @RAMitchell
• Add sample_weight for Ridge (#4696) @lowener
• Unpin dask & distributed for development (#4693) @galipremsagar
• float64 support in treelite->FIL import and Python layer (#4690) @canonizer
• Enable building static libs (#4673) @trxcllnt
• Treeshap hypothesis tests (#4671) @RAMitchell
• float64 support in multi-sum and child_index() (#4648) @canonizer
• Add libcuml-tests package (#4635) @Ethyling
• Random ball cover algorithm for 3D data (#4582) @cjnolet
• Use conda compilers (#4577) @Ethyling
• Build packages using mambabuild (#4542) @Ethyling

? Links

• Development Branch
• Compare with main branch

? Breaking Changes

• Update Python build to scikit-build (#4818) @dantegd
• Bump xgboost to 1.6.0 from 1.5.2 (#4777) @galipremsagar

? Bug Fixes

• Revert "Allow CuPy 11" (#4847) @galipremsagar
• Fix RAFT_NVTX option not set (#4825) @achirkin
• Fix KNN error message. (#4782) @trivialfis
• Update raft pinnings in dev yml files (#4778) @galipremsagar
• Bump xgboost to 1.6.0 from 1.5.2 (#4777) @galipremsagar
• Fixes exception when using predict_proba on fitted Pipeline object with a ColumnTransformer step (#4774) @VamsiTallam95
• Regression errors failing with mixed data type combinations (#4770) @shaswat-indian

? Documentation

• Use common code in python docs and defer js loading (#4852) @galipremsagar
• Centralize common css & js code in docs (#4844) @galipremsagar
• Add ComplementNB to the documentation (#4805) @lowener
• Fix forward-merge branch-22.06 to branch-22.08 (#4789) @divyegala

? New Features

• Update Python build to scikit-build (#4818) @dantegd
• Vectorizers to accept Pandas Series as input (#4811) @shaswat-indian
• Cython wrapper for v-measure (#4785) @shaswat-indian

?️ Improvements

• Pin dask & distributed for release (#4850) @galipremsagar
• Allow CuPy 11 (#4837) @jakirkham
• Remove duplicate adj_to_csr implementation (#4829) @ahendriksen
• Update conda environment files to UCX 1.13.0 (#4813) @pentschev
• Update conda recipes to UCX 1.13.0 (#4809) @pentschev
• Fix #3414: remove naive versions dbscan algorithms (#4804) @ahendriksen
• Accelerate adjacency matrix to CSR conversion for DBSCAN (#4803) @ahendriksen
• Pin max version of cuda-python to 11.7.0 (#4793) @Ethyling
• Allow cosine distance metric in dbscan (#4776) @tarang-jain
• Unpin dask & distributed for development (#4771) @galipremsagar
• Clean up Thrust includes. (#4675) @bdice
• Improvements in feature sampling (#4278) @vinaydes

🚨 Breaking Changes

• Moving more ling prims to raft (#4567) @cjnolet
• Refactor QN solver: pass parameters via a POD struct (#4511) @achirkin

🐛 Bug Fixes

• Fix single-GPU build by separating multi-GPU decomposition utils from single GPU (#4645) @dantegd
• RF: fix stream bug causing performance regressions (#4644) @venkywonka
• XFail test_hinge_loss temporarily (#4621) @lowener
• cuml now supports building non static treelite (#4598) @robertmaynard
• Fix mean_squared_error with cudf series (#4584) @daxiongshu
• Fix for nightly CI tests: Use CUDA_REL variable in gpu build.sh script (#4581) @dantegd
• Fix the TargetEncoder when transforming dataframe/series with custom index (#4578) @daxiongshu
• Removing sign from pca assertions for now. (#4559) @cjnolet
• Fix compatibility of OneHotEncoder fit (#4544) @lowener
• Fix worker streams in OLS-eig executing in an unsafe order (#4539) @achirkin
• Remove xfail from test_hinge_loss (#4504) @Nanthini10
• Fix automerge #4501 (#4502) @dantegd
• Remove classmethod of SimpleImputer (#4439) @lowener

📖 Documentation

• RF: Fix improper documentation in dask-RF (#4666) @venkywonka
• Add doctest (#4618) @lowener
• Fix document layouts in Parameters sections (#4609) @Yosshi999
• Updates to consistency of MNMG PCA/TSVD solvers (docs + code consolidation) (#4556) @cjnolet

🚀 New Features

• Add a dummy argument deep to TargetEncoder.get_params() (#4601) @daxiongshu
• Add Complement Naive Bayes (#4595) @lowener
• Add get_params() to TargetEncoder (#4588) @daxiongshu
• Target Encoder with variance statistics (#4483) @daxiongshu
• Interruptible execution (#4463) @achirkin
• Configurable libcuml++ per algorithm (#4296) @dantegd

🛠️ Improvements

• Adding some prints when hdbscan assertion fails (#4656) @cjnolet
• Temporarily disable new ops-bot functionality (#4652) @ajschmidt8
• Use CPMFindPackage to retrieve cumlprims_mg (#4649) @trxcllnt
• Pin dask & distributed versions (#4647) @galipremsagar
• Remove RAFT MM includes (#4637) @viclafargue
• Add option to build RAFT artifacts statically into libcuml++ (#4633) @dantegd
• Upgrade dask & distributed minimum version (#4632) @galipremsagar
• Add .github/ops-bot.yaml config file (#4630) @ajschmidt8
• Small fixes for certain test failures (#4628) @vinaydes
• Templatizing FIL types to add float64 support (#4625) @canonizer
• Fitsne as default tsne method (#4597) @lowener
• Add get_feature_names to OneHotEncoder (#4596) @viclafargue
• Fix OOM and cudaContext crash in C++ benchmarks (#4594) @RAMitchell
• Using Pyraft and automatically cloning when raft pin changes (#4593) @cjnolet
• Upgrade Treelite to 2.3.0 (#4590) @hcho3
• Sphinx warnings as errors (#4585) @RAMitchell
• Adding missing FAISS license (#4579) @cjnolet
• Add QN solver to ElasticNet and Lasso models (#4576) @achirkin
• Move remaining stats prims to raft (#4568) @cjnolet
• Moving more ling prims to raft (#4567) @cjnolet
• Adding libraft conda dependencies (#4564) @cjnolet
• Fix RF integer overflow (#4563) @RAMitchell
• Add CMake install rules for tests (#4551) @ajschmidt8
• Faster GLM preprocessing by fusing kernels (#4549) @achirkin
• RAFT API updates for lap, label, cluster, and spectral apis (#4548) @cjnolet
• Moving cusparse wrappers to detail API in RAFT. (#4547) @cjnolet
• Unpin max dask and distributed versions (#4546) @galipremsagar
• Kernel density estimation (#4545) @RAMitchell
• Update xgboost version in CI (#4541) @ajschmidt8
• replaces ccache with sccache (#4534) @AyodeAwe
• Remove RAFT memory management (2/2) (#4526) @viclafargue
• Updating RAFT linalg headers (#4515) @divyegala
• Refactor QN solver: pass parameters via a POD struct (#4511) @achirkin
• Kernel ridge regression (#4492) @RAMitchell
• QN solvers: Use different gradient norms for different for different loss functions. (#4491) @achirkin
• RF: Variable binning and other minor refactoring (#4479) @venkywonka
• Rewrite CD solver using more BLAS (#4446) @achirkin
• Add support for sample_weights in LinearRegression (#4428) @lowener
• Nightly automated benchmark (#4414) @viclafargue
• Use FAISS with RMM (#4297) @viclafargue
• Split C++ tests into separate binaries (#4295) @dantegd

? Links

• Development Branch
• Compare with main branch

? Bug Fixes

• Resolve KRR hypothesis test failure (#4761) @RAMitchell
• FIX Accept small floats in RandomForest (#4717) @thomasjpfan
• Remove import of scalar_broadcast_to from stemmer (#4706) @viclafargue
• Replace 22.04.x with 22.06.x in yaml files (#4692) @daxiongshu
• Replace cudf.logical_not with ~ (#4669) @canonizer

? Documentation

• Fix docs builds (#4733) @ajschmidt8
• Change "principals" to "principles" (#4695) @cakiki
• Update pydoc and promote ColumnTransformer out of experimental (#4509) @viclafargue

? New Features

• float64 support in FIL functions (#4655) @canonizer
• float64 support in FIL core (#4646) @canonizer
• Allow "LabelEncoder" to accept cupy and numpy arrays as input. (#4620) @daxiongshu
• MNMG Logistic Regression (dask-glm wrapper) (#3512) @daxiongshu

?️ Improvements

• Pin dask & distributed for release (#4758) @galipremsagar
• Upgrade Treelite to 2.4.0 (#4752) @hcho3
• Simplify recipes (#4749) @Ethyling
• Inference for float64 random forests using FIL (#4739) @canonizer
• MNT Removes unused optim_batch_size from UMAP's docstring (#4732) @thomasjpfan
• Require UCX 1.12.1+ (#4720) @jakirkham
• Allow enabling raft NVTX markers when raft is installed (#4718) @achirkin
• Fix identifier collision (#4716) @viclafargue
• Use raft::span in TreeExplainer (#4714) @hcho3
• Expose simplicial set functions (#4711) @viclafargue
• Refactor tests in cuml (#4703) @galipremsagar
• Use conda to build python packages during GPU tests (#4702) @Ethyling
• Update pinning to allow newer CMake versions. (#4698) @vyasr
• TreeExplainer extensions (#4697) @RAMitchell
• Add sample_weight for Ridge (#4696) @lowener
• Unpin dask & distributed for development (#4693) @galipremsagar
• float64 support in treelite->FIL import and Python layer (#4690) @canonizer
• Enable building static libs (#4673) @trxcllnt
• Treeshap hypothesis tests (#4671) @RAMitchell
• float64 support in multi-sum and child_index() (#4648) @canonizer
• Add libcuml-tests package (#4635) @Ethyling
• Random ball cover algorithm for 3D data (#4582) @cjnolet
• Use conda compilers (#4577) @Ethyling
• Build packages using mambabuild (#4542) @Ethyling

🚨 Breaking Changes

• Move NVTX range helpers to raft (#4445) @achirkin

🐛 Bug Fixes

• Always upload libcuml (#4530) @raydouglass
• Fix RAFT pin to main branch (#4508) @dantegd
• Pin dask & distributed (#4505) @galipremsagar
• Replace use of RMM provided CUDA bindings with CUDA Python (#4499) @shwina
• Dataframe Index as columns in ColumnTransformer (#4481) @viclafargue
• Support compilation with Thrust 1.15 (#4469) @robertmaynard
• fix minor ASAN issues in UMAPAlgo::Optimize::find_params_ab() (#4405) @yitao-li

📖 Documentation

• Remove comment numerical warning (#4408) @viclafargue
• Fix docstring for npermutations in PermutationExplainer (#4402) @hcho3

🚀 New Features

• Combine and expose SVC's support vectors when fitting multi-class data (#4454) @NV-jpt
• Accept fold index for TargetEncoder (#4453) @daxiongshu
• Move NVTX range helpers to raft (#4445) @achirkin

🛠️ Improvements

• Fix packages upload (#4517) @Ethyling
• Testing split fused l2 knn compilation units (#4514) @cjnolet
• Prepare upload scripts for Python 3.7 removal (#4500) @Ethyling
• Renaming macros with their RAFT counterparts (#4496) @divyegala
• Allow CuPy 10 (#4487) @jakirkham
• Upgrade Treelite to 2.2.1 (#4484) @hcho3
• Unpin dask and distributed (#4482) @galipremsagar
• Support categorical splits in in TreeExplainer (#4473) @hcho3
• Remove RAFT memory management (#4468) @viclafargue
• Add missing imports tests (#4452) @Ethyling
• Update CUDA 11.5 conda environment to use 22.02 pinnings. (#4450) @bdice
• Support cuML / scikit-learn RF classifiers in TreeExplainer (#4447) @hcho3
• Remove IncludeCategories from .clang-format (#4438) @codereport
• Simplify perplexity normalization in t-SNE (#4425) @zbjornson
• Unify dense and sparse tests (#4417) @levsnv
• Update ucx-py version on release using rvc (#4411) @Ethyling
• Universal Treelite tree walk function for FIL (#4407) @levsnv
• Update to UCX-Py 0.24 (#4396) @pentschev
• Using sparse public API functions from RAFT (#4389) @cjnolet
• Add a warning to prefer LinearSVM over SVM(kernel='linear') (#4382) @achirkin
• Hiding cusparse deprecation warnings (#4373) @cjnolet
• Unify dense and sparse import in FIL (#4328) @levsnv
• Integrating RAFT handle updates (#4313) @divyegala
• Use RAFT template instantations for distances (#4302) @cjnolet
• RF: code re-organization to enhance build parallelism (#4299) @venkywonka
• Add option to build faiss and treelite shared libs, inherit common dependencies from raft (#4256) @trxcllnt

? Links

• Development Branch
• Compare with main branch

? Breaking Changes

• Moving more ling prims to raft (#4567) @cjnolet
• Refactor QN solver: pass parameters via a POD struct (#4511) @achirkin

? Bug Fixes

• Fix single-GPU build by separating multi-GPU decomposition utils from single GPU (#4645) @dantegd
• RF: fix stream bug causing performance regressions (#4644) @venkywonka
• XFail test_hinge_loss temporarily (#4621) @lowener
• cuml now supports building non static treelite (#4598) @robertmaynard
• Fix mean_squared_error with cudf series (#4584) @daxiongshu
• Fix for nightly CI tests: Use CUDA_REL variable in gpu build.sh script (#4581) @dantegd
• Fix the TargetEncoder when transforming dataframe/series with custom index (#4578) @daxiongshu
• Removing sign from pca assertions for now. (#4559) @cjnolet
• Fix compatibility of OneHotEncoder fit (#4544) @lowener
• Fix worker streams in OLS-eig executing in an unsafe order (#4539) @achirkin
• Remove xfail from test_hinge_loss (#4504) @Nanthini10
• Fix automerge #4501 (#4502) @dantegd
• Remove classmethod of SimpleImputer (#4439) @lowener

? Documentation

• RF: Fix improper documentation in dask-RF (#4666) @venkywonka
• Add doctest (#4618) @lowener
• Fix document layouts in Parameters sections (#4609) @Yosshi999
• Updates to consistency of MNMG PCA/TSVD solvers (docs + code consolidation) (#4556) @cjnolet

? New Features

• Add a dummy argument deep to TargetEncoder.get_params() (#4601) @daxiongshu
• Add Complement Naive Bayes (#4595) @lowener
• Add get_params() to TargetEncoder (#4588) @daxiongshu
• Target Encoder with variance statistics (#4483) @daxiongshu
• Interruptible execution (#4463) @achirkin
• Configurable libcuml++ per algorithm (#4296) @dantegd

?️ Improvements

• Adding some prints when hdbscan assertion fails (#4656) @cjnolet
• Temporarily disable new ops-bot functionality (#4652) @ajschmidt8
• Use CPMFindPackage to retrieve cumlprims_mg (#4649) @trxcllnt
• Pin dask & distributed versions (#4647) @galipremsagar
• Remove RAFT MM includes (#4637) @viclafargue
• Add option to build RAFT artifacts statically into libcuml++ (#4633) @dantegd
• Upgrade dask & distributed minimum version (#4632) @galipremsagar
• Add .github/ops-bot.yaml config file (#4630) @ajschmidt8
• Small fixes for certain test failures (#4628) @vinaydes
• Templatizing FIL types to add float64 support (#4625) @canonizer
• Fitsne as default tsne method (#4597) @lowener
• Add get_feature_names to OneHotEncoder (#4596) @viclafargue
• Fix OOM and cudaContext crash in C++ benchmarks (#4594) @RAMitchell
• Using Pyraft and automatically cloning when raft pin changes (#4593) @cjnolet
• Upgrade Treelite to 2.3.0 (#4590) @hcho3
• Sphinx warnings as errors (#4585) @RAMitchell
• Adding missing FAISS license (#4579) @cjnolet
• Add QN solver to ElasticNet and Lasso models (#4576) @achirkin
• Move remaining stats prims to raft (#4568) @cjnolet
• Moving more ling prims to raft (#4567) @cjnolet
• Adding libraft conda dependencies (#4564) @cjnolet
• Fix RF integer overflow (#4563) @RAMitchell
• Add CMake install rules for tests (#4551) @ajschmidt8
• Faster GLM preprocessing by fusing kernels (#4549) @achirkin
• RAFT API updates for lap, label, cluster, and spectral apis (#4548) @cjnolet
• Moving cusparse wrappers to detail API in RAFT. (#4547) @cjnolet
• Unpin max dask and distributed versions (#4546) @galipremsagar
• Kernel density estimation (#4545) @RAMitchell
• Update xgboost version in CI (#4541) @ajschmidt8
• replaces ccache with sccache (#4534) @AyodeAwe
• Remove RAFT memory management (2/2) (#4526) @viclafargue
• Updating RAFT linalg headers (#4515) @divyegala
• Refactor QN solver: pass parameters via a POD struct (#4511) @achirkin
• Kernel ridge regression (#4492) @RAMitchell
• QN solvers: Use different gradient norms for different for different loss functions. (#4491) @achirkin
• RF: Variable binning and other minor refactoring (#4479) @venkywonka
• Rewrite CD solver using more BLAS (#4446) @achirkin
• Add support for sample_weights in LinearRegression (#4428) @lowener
• Nightly automated benchmark (#4414) @viclafargue
• Use FAISS with RMM (#4297) @viclafargue
• Split C++ tests into separate binaries (#4295) @dantegd

🚨 Breaking Changes

• Fix indexing of PCA to use safer types (#4255) @lowener
• RF: Add Gamma and Inverse Gaussian loss criteria (#4216) @venkywonka
• update RF docs (#4138) @venkywonka

🐛 Bug Fixes

• Update conda recipe to have explicit libcusolver (#4392) @dantegd
• Restore FIL convention of inlining code (#4366) @levsnv
• Fix SVR intercept AttributeError (#4358) @lowener
• Fix is_stable_build logic for CI scripts (#4350) @ajschmidt8
• Temporarily disable rmm devicebuffer in array.py (#4333) @dantegd
• Fix categorical test in python (#4326) @levsnv
• Revert "Merge pull request #4319 from AyodeAwe/branch-21.12" (#4325) @ajschmidt8
• Preserve indexing in methods when applied to DataFrame and Series objects (#4317) @dantegd
• Fix potential CUDA context poison when negative (invalid) categories provided to FIL model (#4314) @levsnv
• Using sparse expanded distances where possible (#4310) @cjnolet
• Fix for mean_squared_error (#4287) @viclafargue
• Fix for Categorical Naive Bayes sparse handling (#4277) @lowener
• Throw an explicit excpetion if the input array is empty in DBSCAN.fit #4273 (#4275) @viktorkovesd
• Fix KernelExplainer returning TypeError for certain input (#4272) @Nanthini10
• Remove most warnings from pytest suite (#4196) @dantegd

📖 Documentation

• Add experimental GPUTreeSHAP to API doc (#4398) @hcho3
• Fix GLM typo on device/host pointer (#4320) @lowener
• update RF docs (#4138) @venkywonka

🚀 New Features

• Add GPUTreeSHAP to cuML explainer module (experimental) (#4351) @hcho3
• Enable training single GPU cuML models using Dask DataFrames and Series (#4300) @ChrisJar
• LinearSVM using QN solvers (#4268) @achirkin
• Add support for exogenous variables to ARIMA (#4221) @Nyrio
• Use opt-in shared memory carveout for FIL (#3759) @levsnv
• Symbolic Regression/Classification C/C++ (#3638) @vimarsh6739

🛠️ Improvements

• Fix Changelog Merge Conflicts for branch-21.12 (#4393) @ajschmidt8
• Pin max dask and distributed to 2012.11.2 (#4390) @galipremsagar
• Fix forward merge #4349 (#4374) @dantegd
• Upgrade clang to 11.1.0 (#4372) @galipremsagar
• Update clang-format version in docs; allow unanchored version string (#4365) @zbjornson
• Add CUDA 11.5 developer environment (#4364) @dantegd
• Fix aliasing violation in t-SNE (#4363) @zbjornson
• Promote FITSNE from experimental (#4361) @lowener
• Fix unnecessary f32/f64 conversions in t-SNE KL calc (#4331) @zbjornson
• Update rapids-cmake version (#4330) @dantegd
• rapids-cmake version update to 21.12 (#4327) @dantegd
• Use compute-sanitizer instead of cuda-memcheck (#4324) @teju85
• Ability to pass fp64 type to cuml benchmarks (#4323) @teju85
• Split treelite fil import from forest object definition (#4306) @levsnv
• update xgboost version (#4301) @msadang
• Accounting for RAFT updates to matrix, stats, and random implementations in detail (#4294) @divyegala
• Update cudf matrix calls for to_numpy and to_cupy (#4293) @dantegd
• Update conda recipes for Enhanced Compatibility effort (#4288) @ajschmidt8
• Increase parallelism from 4 to 8 jobs in CI (#4286) @dantegd
• RAFT distance prims public API update (#4280) @cjnolet
• Update to UCX-Py 0.23 (#4274) @pentschev
• In FIL, clip blocks_per_sm to one wave instead of asserting (#4271) @levsnv
• Update of "Gracefully accept 'n_jobs', a common sklearn parameter, in NearestNeighbors Estimator" (#4267) @NV-jpt
• Improve numerical stability of the Kalman filter for ARIMA (#4259) @Nyrio
• Fix indexing of PCA to use safer types (#4255) @lowener
• Change calculation of ARIMA confidence intervals (#4248) @Nyrio
• Unpin dask & distributed in CI (#4235) @galipremsagar
• RF: Add Gamma and Inverse Gaussian loss criteria (#4216) @venkywonka
• Exposing KL divergence in TSNE (#4208) @viclafargue
• Unify template parameter dispatch for FIL inference and shared memory footprint estimation (#4013) @levsnv