corrected_lasso() with thefamily = "poisson" option. Thanks to Huang Jun fordiscovering this.coef function fromglmnet,due to a breaking change inglmnet.cv_gds which computes cross-validatedgeneralized Dantzig selector.gds andgmus.coef.corrected_lasso,coef.gds,coef.gmu_lasso, andcoef.gmus.print.corrected_lasso,print.cv_corrected_lasso,print.gds,print.gmu_lasso, andprint.gmus.plot.corrected_lasso to also handle the caseof a single regularization parameter.Rglpk on travis CI, so nowthis package is back inImports, andlpSolveAPI is no longer used.Rglpk. Hence, for the development versionRglpk is back inSuggests, together withlpSolveAPI.Rglpk back toImports andremovedlpSolveAPI, as the latest version ofRglpk passes all tests on CRAN, including osX.Fixed random number seed issue which caused test to fail inR-devel.
Internal fix tofit_corrected_lasso andcv_corrected_lasso. Got rid of duplicated code by callingthe functionset_radius.
SinceRglpk does not install automatically on macOS,this package was moved fromImports toSuggests. In addition,lpSolveAPI wasadded toSuggests. This means that the package shouldbuild also on systems that do not haveRglpk, in particularthe versions of macOS on CRAN.
The changes involved adding an optional linear solver fromlpSolveAPI in the functionmusalgorithm().
Added aplot.gds function, which plots the coefficientsestimated byfit_gds.
Internal adjustment, which removed importing of external packagesinto the namespace, and instead specified the functions explicitly using::.
tidyverse has been removed fromSuggests fieldDESCRIPTION.dplyr andtidyr have been added instead.Similarly,library(tidyverse) in the vignette has beenreplaced bylibrary(dplyr),library(tidyr),andlibrary(ggplot2).
hdme is now on CRAN.