v2.0.1

SIRUS v2.0.1

Diff since v2.0.0

Made the feature_importance function more flexible by allowing the function to be called on feature names that are not in the model under consideration.

v2.0.0

SIRUS v2.0.0

Diff since v1.3.4

Some quite big changes in this v2 release. Firstly, confusing terms such as Split and TreePath have been renamed to SubClause and Clause respectively. This is more in line with the usage of these types on the client side. Furthermore, a sorting step was removed (🎉 ) from the algorithm because it wasn't necessary (best code is no code). Next, the plotting functionality has been simplified by exporting more functions. And finally, the root cause for a the performance problem on regression tasks is narrowed down to the random forest implementation. The multiclass classification performs much better now that the lambda hyperparameter was tuned. It turned out that the model is very sensitive to the choice of lambda, and this has been documented at various places. In the benchmarks, SIRUS.jl is now, without finetuning, outperforming the original R algorithm on all but one task.

Merged pull requests:

• Add JOSS paper (#36) (@rikhuijzer)
• [Breaking] Rename confusing terms such as Split (#67) (@rikhuijzer)
• Remove rule sorting step (#68) (@rikhuijzer)
• Add spell checker to repo (#69) (@rikhuijzer)
• Bump actions/checkout from 3 to 4 (#70) (@dependabot[bot])
• Another shot at fixing regression (#71) (@rikhuijzer)
• fix: typo (#73) (@storopoli)
• Extend API (#74) (@rikhuijzer)

Closed issues:

• Add linear models to the benchmarks (#31)
• Easily access condition and consequence (then/otherwise) of a Rule (#44)
• Add API to obtain rules for visualizations (#66)

v1.3.4

SIRUS v1.3.4

Diff since v1.3.3

Merged pull requests:

• Update docs dependencies (#57) (@rikhuijzer)
• Test sirus via RCall (#58) (@rikhuijzer)
• CompatHelper: bump compat for MLJLinearModels to 0.10, (keep existing compat) (#59) (@github-actions[bot])
• add citation of Julia (#63) (@jbytecode)
• Update bibtex (#64) (@jbytecode)
• CompatHelper: add new compat entry for Statistics at version 1, (keep existing compat) (#65) (@github-actions[bot])

Closed issues:

• Performance evaluation may hint at bug? (#51)
• State of the field is missing (#52)
• Code issues in the paper (#54)
• Various doc issues (#55)
• Last details in JOSS paper (#60)

v1.3.3

SIRUS v1.3.3

Diff since v1.3.1

Merged pull requests:

• Improve reproducibility (#53) (@rikhuijzer)
• CompatHelper: bump compat for MLJBase to 1, (keep existing compat) (#56) (@github-actions[bot])

Closed issues:

• Remarks on the binary classification tutorial (#47)
• Inconsistent numerical results (#48)
• Simple example with prediction (#49)
• Comparison with original implementation (#50)

v1.3.1

SIRUS v1.3.1

Diff since v1.3.0

Merged pull requests:

• Switch docs to EvoTrees.jl (#45) (@rikhuijzer)
• Allow Int classes (#46) (@rikhuijzer)

v1.3.0

SIRUS v1.3.0

Diff since v1.2.1

Closed issues:

• Write MLJ-compliant docstrings (#32)
• StableRulesRegressor and StableForestRegressor models should <:Deterministic (#40)

Merged pull requests:

• Add linear models to the benchmarks (#33) (@rikhuijzer)
• Remove LightGBM (#34) (@rikhuijzer)
• Add Breast Cancer Wisconsin dataset (#35) (@rikhuijzer)
• Write MLJ-compliant docstrings (#37) (@rikhuijzer)
• remove deprecated docstring keyword (#38) (@OkonSamuel)
• Add Operations and Fitted parameters sections (#39) (@rikhuijzer)
• Update README.md (#41) (@rikhuijzer)

v1.2.1

SIRUS v1.2.1

Diff since v1.2.0

Closed issues:

• Add test for the MLJ model interface (#29)

Merged pull requests:

• Add tests for the MLJ interface (#30) (@rikhuijzer)

v1.2.0

SIRUS v1.2.0

Diff since v1.1.2

Closed issues:

• Use L2 (Ridge) for determining weights (#7)
• Ensure that weights sum to one (#6)
• Regression is not implemented yet (#13)
• Multiclass-classification is not implemented yet (#15)
• Optimize the rule removal step by filtering flipped versions (#21)

Merged pull requests:

• Update test dependencies (#12) (@rikhuijzer)
• Move cutpoints in separate module (#14) (@rikhuijzer)
• Implement regression (#16) (@rikhuijzer)
• Implement StableForestRegressor (#17) (@rikhuijzer)
• Replace Probabilities by LeafContent and define StableRulesRegressor (#18) (@rikhuijzer)
• Document the rules extraction and refactor (#19) (@rikhuijzer)
• Add a test for a bug in the rule extraction (#20) (@rikhuijzer)
• Make rule removal depend on gap in output (#22) (@rikhuijzer)
• Also sort by rule frequency (#23) (@rikhuijzer)
• Implement and test multiclass-classification (#26) (@rikhuijzer)

v1.1.2

SIRUS v1.1.2

Diff since v1.1.1

Merged pull requests:

• Disable coverage (#11) (@rikhuijzer)

v1.1.1

SIRUS v1.1.1

Diff since v1.1.0

Merged pull requests:

• CompatHelper: bump compat for MLJLinearModels to 0.9, (keep existing compat) (#9) (@github-actions[bot])