0.5.2

What's Changed

• Bump libxgboost-2.0.3-linux-x64 to 1.0.3 by @mdabros in #65
  Fix for the issue mentioned here: #45 (comment)

Full Changelog: v0.5.1...v0.5.2

0.5.1

What's Changed

Note that this release contains a breaking change with regards to missing value handling. Previous default was -1.0F. New default is float.NaN. This change is made to make XBoostSharp more inline with the official XGBoost.

New features

• Add GetFeatureImportance by @mdabros in #48
• Improve Missing Value / null Handling by @MiWeiss in #49

General code and tool improvements

• Bump packages MSTest, Nerdbank.GitVersioning, System.Text.Json, and Microsoft.SourceLink.GitHub by @mdabros in #50
• .editorconfig: Update .editorconfig and add Roslynator.Analyzers/.Formatting.Analyzers by @mdabros in #51
• Code cleanup: DMatrix and Booster by @mdabros in #52
• Dependabot: Add dependabot.yml and enable version updates by @mdabros in #53
• Dependabot: Group Roslynator by @mdabros in #56
• Bump the roslynator group with 2 updates by @dependabot in #60
• Bump codecov/codecov-action from 3 to 5 by @dependabot in #57
• Bump actions/setup-dotnet from 3 to 4 by @dependabot in #58
• Bump actions/checkout from 3 to 4 by @dependabot in #59
• Update native packages for osx-arm64, osx-x64, and linux-x64 by @mdabros in #61
• Bump version to 0.5 before new release by @mdabros in #62

New Contributors

• @dependabot made their first contribution in #60
• @MiWeiss made their first contribution in #49

Full Changelog: v0.4.7...v0.5.1

0.4.8

What's Changed

• Readme: Fix dotnet status and add nuget and downloads by @mdabros in #44
• Add advanced prediction results by @madskonradsen in #39

Full Changelog: v0.4.6...v0.4.8

0.4.6

What's Changed

• Use AnyCPU as solution platform to support other runtimes than x64 by @mdabros in #41

Full Changelog: 0.4.5...v0.4.6

0.4.5

What's Changed

• Add feature type and feature name handling by @madskonradsen in #36

Full Changelog: v0.4.4...v0.4.5

0.4.4

What's Changed

• Readme: Add links to nuget packages by @mdabros in #29
• Use SafeHandle by @mdabros in #31
• Add methods to load and save the models as bytearrays by @madskonradsen in #32

New Contributors

• @madskonradsen made their first contribution in #32

Full Changelog: v0.4.1...v0.4.4

0.4.1

What's Changed

Initial release of XGBoostSharp .net wrapper for XGBoost. This is work based on the now deprecated XGBoost.Net. Initial release has similar features to that wrapper, with a few additions to the supported parameters for the classifier and regressor. Also, XGBoostSharp targets XGBoost 2.0 via libxgboost for both win-x64, linux-x64, and macos-x64. Recommended nuget package to use is XGBoostSharp-cpu, since this includes the references to the native lib. Using the XGBoostSharp package requires separate references to one of the native packages. See README.md for more information. Idea is to start adding features from here to expand XGBoostSharp to have similar features as the official XGBoost python wrapper. This includes separate packages for cuda etc.

• XGBoostSharp package and XGBoostSharp-cpu package (including native references for win-x64, linux-x64, macos-x64)
• Add missing (non-complex) arguments to XGBRegressor and XGBClassifier
• Enable CI and release pipeline for win-x64, linux-x64, macos-x64 using libxgboost 2.0
• Code cleanup
• Target .netstandard2.0
• Initial copy of XGBoost.Net