3.6-0
* Improved efficiency for cox models (avoids recalculating linear
predictors)
* Exported std() function for standardizing a design matrix
* Reorganized testing suite
* Fixed bug in predict.cv.ncvsurv
* Added 'quick start' vignette
* 'survival' package now used for setupLambda in Cox models
3.5-2
* Fixed bug for ncvsurv with integer penalty factors
* Fixed rare numerical accuracy bug in cv fold assignments
* Fixed LOOCV bug introduced by bias-correction feature
* Added user interrupt checking
3.5-1
* Fixed bug with penalty.factor for cv.ncvsurv when some columns may be
degenerate
* Replaced AUC function with more efficient version using survival
package
* Compute bias correction for CV error; this is an experimental feature
at this point and may change in the future
3.5-0
* Substantial efficiency improvements throughout for Cox models.
Coordinate descent redesigned to work in O(n) instead of O(n^2)
operations, and R code redesigned at various points to avoid the
creation of any n x n matrices when fitting and cross-validating Cox
regression models.
* Added function AUC() to calculate cross-validated AUC values for
ncvsurv models.
* Changed method for calculation of cross-validation loss in
cv.ncvsurv().
* More accurate calculation for convexMin in the presence of unpenalized
variables
* Bug fix for factor-valued y with CV logistic regression
* Added option to return fitted values from cross-validation folds
(returnY=TRUE) to cv.ncvreg and cv.ncvsurv.
* Better double/int type checking for penalty.factor
* Modifications to NAMESPACE for compatibility with R 3.3.
3.4-0
* Expanded predict function for Cox models. predict.ncvsurv now
estimates subject-specific survival functions and medians.
* Added plot method for survival curves.
* Added option in perm.ncvreg to permute residuals for linear regression
* On a related note, added permres() function to estimate false
inclusion rates based on residuals at a specific value of lambda
* Added some support for factors in X, y. It is still recommended that
users convert X to a numeric matrix prior to fitting in order to
ensure that predict() methods work properly, but ncvreg will now allow
you to pass a data frame with factors and handle things appropriately.
* Fixed a bug for predict.ncvsurv when applied to models with
saturation issues
* Fixed small memory leak in ncvsurv
3.3-0
* Support for fitting survival models added (ncvsurv), along
with predict, plot, and cv.ncvsurv support functions. Currently,
Cox models are the only type of survival model implemented.
* Added parallel support for cv.ncvreg (with help from Grant Brown)
* Fixed bug in cv.ncvreg when attempting to use leave-one-out
cross-validation (thank you to Cajo ter Braak for pointing this
out)
* ncvreg_fit taken offline; it will return in a future version of
the ncvreg package.
3.2-0
* Updated algorithm to 'hybrid' strong/active cycling
* Automatically coerces X to matrix and y to numeric if possible
* Made ncvreg_fit more user-friendly: user no longer has to
specify lambda, works with coef, predict, plot, etc.
* Fixed bug in convexMin when used with penalty.factor option
* Modified order of arguments for predict so that 'type' comes
before 'lambda' and 'which'.
3.1-0
* Added support for Poisson regression
* Fixed bug in ncvreg_fit that could arise when fitting a model
without an intercept
* Fixed bug in cv.ncvreg with univariate regression (thank you to
Diego Franco Saldana for pointing this out)
3.0-0
* Revised internal algorithms to incorporate targeted cycling
based on strong rules
* Moved standardization to C
* Moved calculation of lambda sequence to C
* As a result of the above three changes, ncvreg now runs
much faster for large p
* Added fir(), perm.ncvreg(), and plot.fir() functions for the
purposes of estimating and displaying false inclusion rates;
these are likely to evolve over the next few months
* Fixed a bug in cv.ncvreg for user-specified lambda sequence
2.7-0
* Added "vars" and "nvars" options to predict function.
* Modified look of summary(cvfit) output.
* Modified internal details of .Call interface.
2.6-0
* Introduction of function ncvreg_fit for programmers who want to
access the internal C routines of ncvreg, bypassing internal
standardization and processing
* Internal restructuring: .Call now used instead of .C
* Bug fix for axis annotations with plot.cv.ncvreg when model is
saturated.
* Bug fix for deviance calculation; would return NaN if fitted
probabilities of 0 or 1 occurred for binomial outcomes.
* Bug fix in NAMESPACE for coef.cv.ncvreg and predict.cv.ncvreg
* Added vertical.line and col options to plot.cv.ncvreg
2.5-0
* Added options in plot.cv.ncvreg to plot estimates of r-squared,
signal-to-noise ratio, scale parameter, and prediction error in
addition to cross-validation error (deviance)
* Added summary method for cv.ncvreg which displays the above
information at lambda.min, the value of lambda minimizing the
cross-validation error.
* Fixed bug in cv.ncvreg with user-defined lambda values.
2.4-0
* Fixed error in definition/calculation of cross-validation
error and standard error
* Added penalty.factor option
* cv.grpreg: Now returns full data fit as well as CV errors
* coef and predict methods now accept lambda as argument
* Added logLik method (which in turn allows AIC/BIC)
* Fixed bug that arose if lambda was scalar (instead of a vector)
* cdfit_ now returns loss (RSS for gaussian, deviance for binomial)
* Fixed bug in cv.ncvreg for linear regression -- cross-validation
was being carried out deterministically
(Thank you to Brenton Kenkel for pointing this out)
* Internal change: standardization more efficient
* Fixed bug: Intercept for logistic regression was not being
calculated for lamda=0
2.3-2
* Fixed formatting error in citation.
2.3-1
* plot.ncvreg: Made the passing of arguments more flexible, so
that user can pass options concerning both the plot and the lines.
* plot.ncvreg: Changed some of the default settings with respect
to color (hcl instead of hsv) and line width.
2.3
* cv.ncvreg.Rd: Fixed the documentation, which no longer agreed
with the function usage. This was an oversight in the release of
version 2.2.
2.2
* cv.ncvreg: Divorced cross-validation from fitting. From a user
perspective, this increases flexibility, although obtaining the
model with CV-chosen regularization parameter now requires two
calls (to ncvreg and cv.ncvreg). The functions, however, are
logically separate and involve entirely separate methods.
* plot.cv.ncvreg: Developed a plotting method specific to
cv.ncvreg objects.