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)
R/llike.R and updated code. (result was
close so not a large effect)\donttest rather
than \dontrunoutput of bark is now a bark object
which will allow S3 methods
replace input arguments y.train and
x.train to use model formula and dataframe as
inputs.
replaced the argument type with two logical
variables: selection and common_lambdas that
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.c and updated
the NAMESPACE
replaced kernel calculation using .C with
.Call to improve speed in
src/kernelCalculationCall.cpp and
R/llike.R
added unit tests in testthat so 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 roxygen
tags
deprecated functions sim.Circle,
sim.Friedman1, sim.Friedman2,
sim.Friedman3 and created new versions
sim_circle, sim_Friedman1,
sim_Friedman2, sim_Friedman3 to 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.cpp and coerce inputs to correct
type (unit tests in testthat/test-llike.R) reported
at
GitHub Issue #3 Addressed error in when p = 1 where
subsetting resulted in output being a vector due to drop in dimension.
Added unit test in testthat/test-bark.R