NEWS.md

In abess: Fast Best Subset Selection

abess 0.4.10

• Fix note in NOTE about possible bashisms.

abess 0.4.9

• Fix bug in Cpp level
• Fix error in: https://www.stats.ox.ac.uk/pub/bdr/clang19/abess.log
• Fix notes in https://cran.r-project.org/web/checks/check_results_abess.html

abess 0.4.8

• Support no-intercept GLM model by param 'fit.intercept'.
• Allow to restrict the range of estimation for beta by param 'beta.high' and 'beta.low'.
• Add cite message when load 'abess'.
• Fix a bug when support.size is 0.

abess 0.4.7

• Allow the other criterion for model selection: AUC for (multinomial) logistic regression such as the area under the curve (AUC).
• Simplify the C++ code structure.
• Fix note "Specified C++11: please update to current default of C++17" in CRAN.

abess 0.4.6

• Adapt to the API change of the Matrix package.
• Change the package structure such that the API functions can reuse the utility function. It facilitates the testing for package.
• Update citation information.

abess 0.4.5

• Support generalized linear model for ordinal response, also named as rank learning in machine learning community.
• Support robust principal analysis
• Modify R package structure to make many internal components are reusable.
• Update README.md

abess 0.4.0

• Support generalized linear model when the link function is Gamma distribution. By setting family = "gamma" in abess function, users can analyze the dataset with a positive valued and skewed response.
• Support flexible support size for sequential principal component analysis (PCA), accompanied with several helpful generic function like plot.
• Support user-specified cross validation division for abess and abesspca function by additional argument foldid.
• Support robust principal component analysis now. A new R function abessrpca can access it.
• Improve the R package document by: adding more details and giving more links related to core functions.

abess 0.3.0

• Add docs2search for R's website
• Support important searching to improve computational efficiency when dimension is 10,000.

abess 0.2.0

• Support sparse matrix as input
• Support golden section search for optimal support size
• Support ridge-regularized penalty as a generic component
• Support group subset selection as a generic component
• Best subset selection for principal component analysis via abesspca
• Bug fixed

abess 0.1.0

• Initial stable version abess package
• Support best subset selection for linear regression, logistic regression, poisson regression, cox proportional hazard regression, multi-gaussian regression, multi-nominal regression.
• Support nuisance selection as a generic component
• Unified API for cross validation and information criterion to select the optimal support size.
• A documentation website is support for abess package

abess documentation built on April 11, 2025, 6:09 p.m.