The hardware and bandwidth for this mirror is donated by METANET, the Webhosting and Full Service-Cloud Provider.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]metanet.ch.
betareg models: confidence intervals
for the mean submodel are computed using a delta-method
approximation.survival::coxph(). Conditional effects are plotted on the
linear predictor (log-hazard) scale. Raw survival data (event times and
censoring) are not plotted, and confidence intervals are based on
standard errors of the linear predictor.by_breaks argument: allows users to manually
specify conditioning values when by is numeric, replacing
the previous fixed default behavior. If not provided, numeric
by variables with more than 6 unique values default to
their 10th, 50th, and 90th percentiles.rlm_vcov argument: robust standard errors for
MASS::rlm() models can now be computed using any sandwich
estimator type (e.g., "HC3", "HC5",
"const", etc.), a user-supplied covariance matrix, or a
custom function.ci_polygon_alpha argument: allows users to control
the transparency of polygon-style confidence interval bands when
ci_type = "polygon". Defaults to 0.5 and accepts values
between 0 and 1.show_conditioning argument: allows users to print a
concise summary in the console describing how predictions are
conditioned. When show_conditioning = TRUE,
easyViz() reports which variables are held fixed and which
vary in the prediction grid, whether predictions from mixed-effects or
GAM models are conditional on random effects or marginal, and, for
numeric by variables, the values at which predictions are
evaluated (e.g., quantiles or user-specified by_breaks).
This option does not affect plots or returned values and is intended as
a diagnostic aid.plot argument: allows users to suppress graphical
output. When plot = FALSE, easyViz() returns
only the predictions (as an invisible easyviz.pred.df
object) without drawing a plot. Useful when storing or further
processing predicted values.offset = are now automatically detected. For
log-link count models (e.g., Poisson, negative binomial, COM-Poisson,
generalized or truncated count models), observed responses are
automatically scaled by the offset for plotting, and the prediction grid
sets the offset to 1 by default, ensuring predictions and raw data are
displayed on a common rate scale.by variables: numeric
conditioning variables with a small number of distinct values
(currently ≤ 6) are now treated as discrete groups and plotted with one
line per value (like a factor), instead of always using quantile
cross-sections. For truly continuous numeric by variables
(more than 6 unique values), easyViz() still uses default
cross-sections (10th, 50th, and 90th percentiles) or the user-supplied
values from by_breaks. This makes the behavior more
intuitive for common cases such as 0/1 or small ordinal scales.by variable is
specified and users do not explicitly set add_legend,
easyViz() now automatically enables the legend. This
ensures that multi-line or multi-group plots always display a clear
legend by default, while preserving user control when
add_legend is manually set."out" for
legend_position: setting
legend_position = "out" applies a predefined layout that
places the legend outside the plotting region above the panel.re.form to re_form for consistency
with other arguments. Argument re.form is deprecated.
Please use re_form instead.ci_level argument to control the confidence
interval level (default is 0.95). The confidence level used for the
interval is included in the returned prediction data frame.pred_resolution argument,
which controls the number of prediction points used for numeric
predictors. The default is 101 (matching
visreg), but higher values can be helpful for predictors
with a wide range or when visualizing nonlinear relationships.ci_type = NULL to suppress confidence
intervals for numeric predictors.legend_title argument.legend_title_size argument.legend_horiz argument.legend_args argument, which allows passing any additional
parameters to legend() (e.g., pt.cex,
bg, box.lwd, inset).legend_text_size to
legend_label_size for consistency with
legend_title_size and legend_labels. The
argument legend_text_size is deprecated. Please use
legend_label_size instead.plot_args (e.g.,
cex.axis, col.axis, font.axis)
when the predictor is a factor.lm(), rlm(), gls(), and
nls() models. For models with asymptotic inference (e.g.,
glm(), mgcv::gam(), glmer(),
glmmTMB()), confidence intervals are based on standard
normal approximations.by is specified for
interactions, all prediction lines are now drawn on top of their
associated confidence intervals, ensuring they remain visible and are
not obscured by overlapping CI polygons.plogis() for greater numerical stability, and the
cloglog inverse is now more robust to extreme values.point_col) have a length mismatch with the cleaned
data used for predictions, helping users avoid misaligned plot
aesthetics when the original data contains missing values.by and fix_values arguments in a
multi-panel layout.These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.