NEWS

glmnetUtils 1.1.9 (2023-09-10)

Skip some tests on 32-bit Solaris R-patched due to numerical convergence issues.

Add some plotting options for plot.cva.glmnet: log.x controls whether to plot the X-axis (lambda) on the log scale, and the legend can be omitted by setting either legend.x or legend.y to NULL.
Compatibility fixes for glmnet 4.1-1.

Fix handling of non-factor categorical predictors (from R 4.0, data frames will not have character columns converted to factors by default). The practical impact of this should be minor.

Support relaxed (non-regularised) fits in glmnet.formula and cv.glmnet.formula (requires glmnet 3.0 or later).
Add a legend when plotting a cva.glmnet object.

Fixes a bug in the assignment of observations to crossvalidation folds in cva.glmnet. The impact is most serious for small datasets, where the number of observations per fold is relatively low. If you are using this function, it's highly recommended you update the package.

Now allows interaction and expression terms without requiring use.model.frame=TRUE. This works in an additive fashion, ie the formula ~ a + b:c + d*e is treated as consisting of three terms, a, b:c and d*e each of which is processed independently of the others. A dot in the formula includes all main effect terms, ie ~ . + a:b + f(x) expands to ~ a + b + x + a:b + f(x) (assuming a, b and x are the only columns in the data). Note that a formula like ~ (a + b) + (c + d) will be treated as two terms, a + b and c + d.
The call component of a glmnet/cv.glmnet object that uses the original matrix/vector interface is now useful.
You can now explicitly specify the vector of crossvalidation folds (for the inner loop over lambda) when calling cva.glmnet.
Correctly handle non-syntactic factor variables in a formula.