NEWS.md
Fix mishandling of response variables with a robust scale of 0, e.g., 0-inflated responses or responses with more than 50% identical values.
adapense()
and adapense_cv()
).pense()
, adapense()
, regmest()
, etc.) are not estimating prediction performance via cross-validation anymore. This can now be done using the corresponding functions pense_cv()
, adapense_cv()
, and so on.prediction_performance()
to summarize the prediction performance of several fits.plot()
, coef()
, summary()
, and predict()
methods for cross-validated fits also implement the “one-standard-error rule” (with the “1” adjustable by the user).en_admm_options()
.correct
in pense()
, pensem()
, coef()
, etc., is not supported anymore and will be ignored with a warning. All estimates are now uncorrected (i.e., correct=FALSE
in previous versions of the package).pensem()
is now called pensem_cv()
.initest_options()
is replaced by enpy_options()
using better naming of arguments.en_options_aug_lars()
and en_options_dal()
are replaced by en_lars_options()
and en_dal_options()
for more consistent naming.pense_options()
and mstep_options()
are superseded by mm_algorithm_options()
and arguments specified in the calls to pense()
and companions.enpy()
is replaced by enpy_initial_estimates()
which has different default argument values.Deprecated functions can still be used (for now) with a warning.
STRICT_R_HEADERS
)pense
and pensem
. Note: The lambda values in this release are not the same as in previous releases!Fixed a bug when computing “exact” principal sensitivity components.
predict()
function for pensem
objects if computed from a fitted pense
object.delta
and cc
specified in pense_options()
for the initial estimator. Remove delta
and cc
arguments from initest_options()
and instead add them to enpy()
.Add further measure of the prediction performance (resid_size
) to obj$cv_lambda_grid
, where obj
is of class pense
or pensem
.
Initial stable release of the package.