fitpgam() and related functions added to fit a
log-additive model over two time scales.fitvcm() and related functions added to fit a
varying-coefficient model over two time scales.select_model2ts() added to compare several fitted
models and identify the best-fitting one.make_grid() facilitates the creation of a new plotting
grid.prepare_data(),
fit1ts(), and fit2ts().predict.haz2ts() can now predict also including
covariates fixed at arbitrary values
GLAM_2d_covariates() returns covariances between the
alpha and beta parameters
All functions that use ucminf to minimize the AIC/BIC of the
model wrt the smoothing parameter(s) now have a smaller value for the
option xtol. Additionally, it can also be changed in the
control lists.
Fixed a small typo in plot_haz1ts() that did not
allow to plot confidence intervals in color specified by user.
Fixed problem with variable names in
prepare_data_LMMsolver()
print.data2ts() now prints rounded values for total
exposure time
predict.haz2ts() method added to objects of class
'haz2ts'. It allows prediction of the hazard, its standard
errors, the cumulative hazard and the survival probability, from a
fitted model of type 'haz2ts', for arbitrary values of the
time scales. It can also be used to obtain individual predictions for
the original data points.
predict_comprisk2ts() allows prediction for
competing risks models. It takes as input a list of cause-specific
hazard models over two time scales, all fitted with
fit2ts(), on the same grid, and a new data.frame with
values of the two time scales for which predictions are
requested.
DESCRIPTION has been updated, to correct a small
typo.