gp_axis argument to
fix URLs in JSS references.
specify the number of digits of p-values.
bugfix: highlighting was wrong since 1.4-5. It
now correctly recycles the fill parameter along last
dimension (i.e., the
dependent variable), in contrast to standard behavior of
bugfix: doubledecker plot had wrong shading since 1.4-5. It now correctly recycles graphical parameters along last dimension (i.e., the dependent variable), in contrast to standard behavior of mosaic().
small internal change in
tile() to account for new grid
graphical parameters are now recycled along first dimension.
labeling_cells() could not handle structable objects
when called separately from
ternaryplot() now has additional
labels_rot= arguments for changing the angle of the labels.
The loddratio plot now suppresses labels in the 2x2x(strata) case.
Remove outdated inst/doc directory.
lines() method for
gpar() argment to control confidence intervals in
add imports for functions in recommended packages to NAMESPACE
shading_hcl now also uses solid line type for
abs(residuals) < eps like
Add residual-based shading to
residuals() method for
grid_abline() for convenience.
Add P-values to the output print.Kappa() produces.
Fix legend of
distplot() in case of leveled plots.
cotab_agreementplot panel function for plotting
conditional (stratified) agreement charts added.
loddsratio added as an extension of
conditioned generalized odds ratios. The plot method is greatly
enhanced, and structural zeros (missing values) are also
oddsratio is now just an alias for
Bug fix in
"conditional" type, tables are now reordered so that the
conditioning variables come first.
cotab_loddsratio panel function for plotting
conditional log-odds ratios plots added.
assocstats now handles strata (all but the first two
sieve now accepts a
binreg_plot() added for (conditioned) binary regression plots
Bug fix in
condvars= now not only reorders the table, but also the split
information, so that both remain consistent.
All plot functions now have an option to return the produced
plot as a grid object for further use (e.g., in
mplot() added for combining multiple grid plots in a
legend_grid now allows more options for positioning,
inset argument has been added for relative adjustments.
to control plotting of the OLS and WLS lines.
distplot() gets an
Changed default line types for
sieve so that positive residuals are
shown with solid lines, as per Friendly specifications.
fixed problems with
pairs_strucplot() type argument, particularly for
Fixed error in
CoalMiners data (missing group, labels
Change default area type for sieve plots to "area"
several namespace issues fixed
Bug fixed in
grid_legend() extended for better finetuning of graphical parameters
legend_resbased() better handles spacing for labels.
now allow changing the font family.
default diagonal panel in
pairs() is now set to
pairs_diagonal_mosaic(), with alternating labels and
frequencies shown on the bars.
labeling is more customizable in
agreementplot() now allows to add marginals to the plot
abbreviate argument of
labeling_border() renamed to
abbreviate_labs to prevent name clash with
abbreviate_foo args in
Several partial matches fixed in code
assoc() would not plot tables with 0 residuals
structable() adds dimnames and dimname names if none are specified
Bug fix: print error message when subsetting/selecting of structable objects using more than 2 indices
‘NEWS’ file changed to .Rd format
gamma argument removed from
Add aperm method for structable objects
For use with
now sets the line type of borders corresponding to
< eps to
lty in addition to setting
fourfold(), modified default
non-significant log odds ratios to be more visually distinct from
the fully saturated
colors[5:6] for significant ones.
fourfold(), allow the function to work with
tables with more than 3 dimensions,
by restructuring all strata dimensions into
a single combined 3rd dimension.
fourfold(), modified defaults for
mfcol to give landscape display,
nr <= nc,
nr >= nc. If
length(dim(x)) > 3, set
Ord_plot() for devices where the default filling is
"white" and not
"transparent" by explicitly setting it to
Bug fix in
as.table.structable(), returning the table in a different order
than defined in the structable object, confusing in particular
add parameter to
ternaryplot() to control the positioning of the
Small bug fixes in handling of some graphical parameters.
Corrected df handling in
goodfit() with ML estimation:
only non-zero cells are used. This is backward compatible
Fixed bug in
goodfit() for binomial distribution with
"size" parameter (introduced in 1.2-5).
Small typo in doc of
Bundesliga data set has been augmented with the results of the seasons 2006/7, 2007/8, 2008/9 (thanks to Torsten Hothorn).
goodfit() was modified to treat zero cells better:
zero cells (i.e., counts below the maximal observed count that
did not occur in the sample) are retained (and not dropped as
before). ‘Trailing’ zero cells (i.e., counts above the maximal
observed count) are still not considered. The documentation now
points out the problems with the minimum-chi-squared method in
the latter situation.
sieve() now accepts a
gp_tile argument to control the appearance of
the cells (apart from the sieve color)
Bug fix: labeling arguments were incorrectly handled when the options were not provided as named vector
ternaryplot() now makes use of the
cex argument also for
the rendering of optional labels, if any
‘hcl-colors.pdf’ removed from source ball (vignette now in colorspace)
strucplot() now accepts a
df argument that is
passed to the shading functions. Also, expected values are no
longer computed if residuals are given.
Fixed a bug in labeling (incorrect handling of some parameters)