-
improve computational efficiency by saving old log likelihood values rather than recomputing; should improve computational speed by close to a factor of 2
-
replace .C with .Call for computing the design matrix (resolves CRAN warning about noRemap)
- fixed issues identified on CRAN package check regarding noRemap (issue #9)
- identified an error in the calculation of the covariance matrix for the coefficients in
R/llike.Rand updated code. (result was close so not a large effect)
- archived on 3/31 due to unit test failure on MKL with R-devel gcc skip this unit test on CRAN as test is for code that will be deprecated shortly.
- failed MKL checks
- failed MKL checks
- update longer running examples to use
\donttestrather than\dontrun - add reference to DESCRIPTION
-
output of
barkis now abarkobject which will allow S3 methods -
replace input arguments
y.trainandx.trainto use model formula and dataframe as inputs. -
replaced the argument
typewith two logical variables:selectionandcommon_lambdasthat are more intuitive. -
changed running times of examples to address issue https://CRAN.R-project.org/package=bark which led package to being archived in 2015
-
added registration of native routines in foreign function calls and disabled symbol search in
src/bark-init.cand updated theNAMESPACE -
replaced kernel calculation using
.Cwith.Callto improve speed insrc/kernelCalculationCall.cppandR/llike.R -
added unit tests in
testthatso that code coverage is reported with CI; code coverage badge added to README. Unit tests now cover over 99% of the code as reported by CodeCov -
added GitHub actions for CI and checks on Windows, MacOSX, and Ubuntu and added R CMD check passing Badge to README.md in GitHub repo.
-
converted all help files to use
roxygentags -
deprecated functions
sim.Circle,sim.Friedman1,sim.Friedman2,sim.Friedman3and created new versionssim_circle,sim_Friedman1,sim_Friedman2,sim_Friedman3to avoid confusion with S3 methods -
updated CODE_OF_CONDUCT.md, SECURITY.md, CONTRIBUTING.md on GitHub repo and other updates for OpenSSF BestPractices Badge (added to README.md)
-
GitHub Issue #1 Added type checks to
src/kernelCalculationCall.cppand coerce inputs to correct type (unit tests intestthat/test-llike.R) reported at -
GitHub Issue #3 Addressed error in when
p = 1where subsetting resulted in output being a vector due to drop in dimension. Added unit test intestthat/test-bark.R