CRAN release: 2022-05-09
- Added a
NEWS.mdfile to track changes to the package.
formhas been renamed to
formulato reflect common naming conventions in R statistical modeling packages.
coef()methods are now implemented using
future.apply::future_lapplyallowing for computing predictions and extracting coefficients in parallel. This may result in slightly worse performance (due to memory overhead) when the model/prediction data is small but will be significantly faster when e.g. generating predictions on a very large data-set.
- Solved an issue with
formula. The user-supplied value of
formulais handled via
deparse()which has a default
width.cutoffvalue of 60. This was causing issues with formulas by splitting them into multi-element character vectors. It has now been set to the maximum value of
500Lwhich will correctly parse all lengths of formulas.
predict()now forces evaluation of the
formulaargument in the
bolasso()call. This resolves an issue where, if a user passes a formula via a variable,
predict()would pass the variable name to the underlying prediction function as opposed to the actual formula.