glm whose dispersion is
estimated (Gaussian, Gamma, inverse Gaussian and quasi families) use the
t distribution, as summary.glm() does, where they used z
and were too narrow. They widen wherever depictr reports or draws
them.coefficient_plot(standardise = TRUE), and so
model_report(), scales continuous inputs by their standard
deviation and an interaction through its continuous inputs, leaving
factor, logical and two-valued inputs unscaled (Gelman, 2008), where it
scaled every model-matrix column, so a contrast depended on group
balance.coefficient_plot(standardise = TRUE) leaves
poly(), ns() and other matrix terms unscaled,
with a message, and warns about an interaction of continuous inputs that
are not all centred. The axis and panel titles say which inputs were
scaled.coefficient_plot() reads an unnamed labels
vector in model order, whatever order is, where it matched
the vector to the terms as reversed or sorted for display, so a vector
written for 0.3.0 can put labels on other terms. Named labels are
unaffected.model_fit_table() and the model_report()
subtitle give every glm the deviance R², where they applied
McFadden’s ratio to every family and could refit the null model on other
rows. Binary logistic values change only where that refit went
wrong.model_fit_table() takes the AIC, BIC and log-likelihood
of an lme4::lmer() REML fit from a maximum-likelihood
refit, with a message, where it reported REML criteria, which cannot be
compared between models with different fixed effects.vif_plot() computes VIFs from the coefficients’
covariance, as car::vif() does, where it ignored the fit’s
weights. Values change for a weighted lm, a binomial or
Poisson glm and any other glm that weights its
observations unequally. A model without an intercept changes too, with a
warning.tidy_estimates() exponentiates the estimates and limits
of an lm, a glm, an lme4 fit and a data frame
when given exponentiate = TRUE, where it ignored the
argument for these and returned the link scale.effects_plot(), interaction_plot() and
model_report() build the prediction grid from the raw
variables, so predict() re-applies every transformation,
where a transformed term beside its raw variable gave a flat effect line
or coincident interaction lines.calibration_plot() and the Q-Q panel of
residual_diagnostics_plot() pair each outcome of a
glm fitted with na.action = na.exclude with
its own fitted value, where they paired it with another row’s.residual_diagnostics_plot() and
influence_plot() leave out the rows that
na.action = na.exclude excludes, for an lm as
for a glm, without the warning they gave, and the 2p/n
leverage line of influence_plot() no longer counts
them.residual_diagnostics_plot() scales the
quantile residuals of a weighted Gaussian or Gamma model by its prior
weights, which it ignored, and leaves out observations with zero prior
weight.binned_residual_plot() and the binned panel of
residual_diagnostics_plot() draw the band from the fitted
model, where the empirical band was too narrow in bins without events,
and leave out observations with zero prior weight.
binned_residual_plot(band = "empirical"), though not the
panel, restores the old band.model_report() draws binned-residual and
quantile-residual panels for a glm, as
residual_diagnostics_plot() does, where it drew
normal-theory panels that flag a correct logistic model.
glm_panels = FALSE restores the old panels, and
seed fixes the randomised residuals.frequentist_bayesian_plot() and
posterior_plot() keep only the b_ parameters
of a posterior draws object that has them, as one made from
a brms fit does, with a message, where
frequentist_bayesian_plot() drew the group-level effects as
parameters and pooled b_Intercept with
Intercept.frequentist_bayesian_plot() and
posterior_plot() leave out the group-level effects of a
stanreg fit and the log densities lp__ and
lprior of any draws, which they drew as parameters.frequentist_bayesian_plot() with draws plots only the
terms both inputs share, with a message naming those it drops, where
sigma and terms without a posterior took rows of their own,
and sets the outer posterior interval by conf_level, where
it stayed at 95%.frequentist_bayesian_plot() reads a
brms::fixef() matrix as a summary, where it drew the
matrix’s four columns as parameters.survival_plot() draws log-log confidence limits by
default (Borgan & Liestøl, 1990), where it drew untransformed
Greenwood limits. Unless time is a survfit
object, conf_type = "plain" restores them, apart from the
last step of a curve that falls to zero. conf_type = "log"
gives the default of survival::survfit().survival_plot() recomputes a survfit
object’s limits at conf_level and conf_type,
where it copied the object’s own and ignored conf_level.
conf_type = "log" restores the limits of an object made
with the defaults of survfit().survival_plot() draws limits for a survfit
object made with conf.type = "none", where it drew none.
conf_level = NA leaves them out. An object without standard
errors, or with Peto or modified lower limits, keeps its own and warns
when they differ from those asked for.survival_plot() reads the status coding
once over all observations, where it read each group separately and so
drew a group whose subjects are all censored under the 1/2 coding
falling to zero. That group’s curve now stays at one.survival_plot() honours group and
status, as a column name or a vector, when
time is a data frame, where it read them only from
conventionally named columns and could pool the groups.survival_plot() counts a
subject followed up to exactly a break and keeps a last break equal to
the longest follow-up, both of which a floating-point error in the
default breaks left out.survival_plot() follow
survival::quantile.survfit(), where a rounding error could
label the wrong time, a survfit curve flat at 0.5 could
take its midpoint from a censoring time and a curve that ends at 0.5 got
no guide.time is a survfit object, the
confidence band of survival_plot() reaches the last event
of a curve that falls to zero, where it stopped one event early.k_diagnostic() and cluster_plot() default
to k_range = NULL, which searches 1:8 for the
gap and elbow criteria. The old 2:8 let the gap statistic
find clusters in pure noise and kept the elbow from suggesting two.
k_range = 2:8 restores it.silhouette_plot() keeps the cluster labels it is given
in its bands and in its "silhouette" table, whose
cluster and neighbor columns are factors of
them, where it recoded them to 1, 2, 3 and so on and gave
neighbor as integer codes.ts_forecast() and the forecast overlay of
timeseries_plot() forecast the seasonally adjusted series
with a random walk with drift, as
forecast::stlf(method = "rwdrift", s.window = "periodic")
does, where they extrapolated the last cycle’s trend with intervals far
too narrow. The overlay caption names the method.timeseries_plot() continues a forecast along the
history’s own axis, numeric or Date, where it restarted
near the start of the axis, and sorts an unsorted time
before forecasting.timeseries_plot() centres the moving average of an even
rolling window, as a 2×12 average for
rolling = 12 (Hyndman & Athanasopoulos, 2021), where
the average was drawn half an observation early. Odd windows are
unchanged.timeseries_plot() draws one line per column of a
multivariate ts or a matrix, where it joined the columns
into one series.summary_table() summarises a date or date-time column
as one Range row and a duration as mean (SD) in a unit that
the statistic names, where it gave one row per distinct value.palette_safety() and check_figure()
measure colour separability with CIEDE2000 (Sharma, Wu & Dalal,
2005), where the 1976 distance overstated the differences between
saturated colours, so distances and some verdicts change.
metric = "cie76" restores the old measure.simulate_cvd() at partial severity interpolates between
the published matrices of Machado et al. (2009), where it drew the
full-severity matrix towards the identity, so its colours between
severities 0 and 1 change. Full severity is unchanged.check_figure() measures each text against its own
background, holds large text to 3:1 and measures a translucent plot or
panel background as the colour it shows, where it measured all text
against the plot background at 4.5:1.check_figure() composites translucent colours over
their background with the alpha each geom applies, where it measured
them as opaque, and compares colours and fill colours each among
themselves, where it pooled them, so verdicts can change in either
direction.check_figure() counts a colour or fill split across
layers, one value per layer, as a colour encoding, where it reported
none, and counts shape and line type split that way towards
redundant_encoding.check_figure() adds a tenth row,
annotation_text_size, which holds the text inside the
panel, such as annotations, geom_text() labels and the cell
values of depictr’s plots, to the size floor, so a figure that passed
can fail.coefficient_plot(), timeseries_plot() and
decompose_plot() take new arguments before
title, palette and confidence
respectively, so those arguments and the ones after them must be passed
by name.tidy_estimates(), save_plot(),
frequentist_bayesian_plot(),
scale_colour_depictr() and
scale_fill_depictr() take a new argument before
..., so an argument meant for ... must be
passed by name.conf_level outside (0, 1), such as 95, is refused by
every function that takes one, where a level of 0 or 1 gave the model
estimates zero-width or unbounded intervals, 95 gave missing intervals
or an error and roc_curve_plot() without ci
ignored the level. Without ci,
roc_curve_plot() also refuses NULL, which it
ignored.conf_level, except
survival_plot() and calibration_plot(),
refuses NA, where most gave missing intervals. Those two
still take it to leave the limits out, and both refuse NaN,
which they read as NA.compare_models() and
frequentist_bayesian_plot() read an unnamed
labels vector as one label per term, in the order the terms
first appear, where they took one label per row of the stacked estimates
or per posterior draw, and refuse a vector of the old length.posterior draws object that holds b_
parameters, frequentist_bayesian_plot() and
posterior_plot() take labels keys without the
prefix, so a key such as b_rainfall written for 0.3.0 no
longer matches.survival_plot() refuses a factor, character or
non-integer status, which it read by its level codes, as
missing or truncated to an integer.survival_plot() refuses a status coded 0/1
in some groups and 1/2 in others, which it read group by group.survival_plot() refuses a group or
status name that matches no column of a data frame
time, which it ignored.effects_plot(), interaction_plot() and
model_report() refuse a formula term computed from
statistics of its rows, such as I(x - mean(x)), which
shifted the line, a formula that reads columns with $ and a
transformed term whose data the call cannot supply.frequentist_bayesian_plot() with draws refuses inputs
that share no terms and two terms of either input that would share a
label, and posterior_plot() refuses a
brms::fixef() matrix and two parameters that would share a
label.explore_distribution(), ecdf_plot(),
raincloud_plot(), group_comparison_plot(),
explore_categorical(), explore_pairs(),
ridgeline_plot() and outlier_plot() refuse a
numeric group with more than 20 distinct values, which they
drew, and scatter_trend() refuses one when called, where it
returned a plot that failed when drawn.scatter_trend(), explore_distribution(),
ecdf_plot(), raincloud_plot(),
explore_categorical() and outlier_plot()
refuse a group missing for every row when called, where
they drew it or returned a plot that failed when drawn.power_curve_plot() refuses rows of one curve that share
a sample size, where it joined them into a zigzag line.acf_plot(), decompose_plot(),
seasonal_plot() and ts_forecast() refuse a
series of more than one column, and timeseries_plot()
refuses a forecast for one, where they joined the columns
into one series.timeseries_plot() refuses an even rolling
window as long as the shortest series, since the centred average spans
one observation more than the window.pca_plot() refuses a misspelt group
column, a group vector of the wrong length and a column
name passed with a prcomp object, which it read as one
level or padded with NA.coefficient_plot(), compare_models(),
frequentist_bayesian_plot(),
optimizer_fixef_plot() and the coefficient panel of
model_report() label a two-level factor or logical input
with its level, as in treatment: enhanced, where they
showed treatment. Except in model_report(),
labels = c(treatmentenhanced = "treatment") restores an old
label.rainfall × treatment: enhanced, where they showed
rainfall × treatmentenhanced. Except in
model_report(), a labels entry for the raw
term restores it.compare_models() labels the factor levels of every
model, as in region: South, where it labelled only those of
the first model holding a factor and showed the rest as
regionSouth. A labels entry for the raw term
restores them.compare_models(x_lab = NULL), and
frequentist_bayesian_plot(x_lab = NULL) with summaries,
draw the default axis title, ‘Estimate’ or the ratio’s name, where they
drew none. Adding ggplot2::labs(x = NULL) removes it.model_report() subtitle of a glm calls
its R² ‘deviance R²’, where it said ‘pseudo-R²’.estimation_plot() and
group_comparison_plot(differences = TRUE) give every group
a column in the difference panel, the reference’s first and empty, so
each difference sits in its own group’s column, where the difference
panel had one column fewer than the groups and spread across the full
width.estimation_plot() and
group_comparison_plot(differences = TRUE) writes a round
number of resamples in full, as 100,000, where it printed 1e+05.group_comparison_plot(differences = TRUE) keeps each
group’s colour whatever the reference, where the colours
moved between groups, and estimation_plot() matches a named
palette to the levels by name.frequentist_bayesian_plot() draws the frequentist
estimates in brand blue and the Bayesian ones in orange on both paths,
where the draws path swapped them, and adds a caption naming each
source’s interval level. caption = NA omits it.save_plot() writes PDF, EPS and PostScript files with
the cairo devices, which embed the fonts, and a TIFF LZW-compressed on
white unless bg is given, where the files failed journals’
checks. device = "pdf", "eps",
"ps" or "tiff" restores ggplot2’s
devices.save_plot() rasterises the translucent layers of an EPS
or PostScript file at 600 dpi or more, with a message, where it left
them out.save_plot()
writes a PDF, EPS or PostScript file with a message or warning saying
what the file lacks, and a Mac with neither XQuartz nor ragg writes the
TIFF uncompressed, with a warning.seasonal_plot(style = "season") draws its cycle lines
from a ramp that starts at #4a8fc0, at 3.51:1 against
white, where the earliest cycles were close to invisible.
palette = depictr_palette(n, type = "sequential"), with
n the number of cycles, restores the old ramp.palette_preview(), confusion_matrix_plot()
and correlation_heatmap() draw each label in black or
white, whichever contrasts more with its tile, where some labels fell
below the 4.5:1 that WCAG asks of text.pca_plot() fixes the aspect ratio, so a figure of
another shape gains blank space, and draws its loading arrows and labels
in dark grey, where they took the accent, the palette’s fourth
colour.group, power_curve_plot() draws
one curve per level of an effect, param or
term column, with a message, where it joined them into one
line.power_curve_plot() draws a confidence band from
<power>_lo and <power>_hi columns,
such as the power_lo and power_hi of a pilotr
sweep over sample size, where it drew none, and reads them before
lower and upper.scatter_trend(), explore_distribution(),
ecdf_plot(), raincloud_plot(),
group_comparison_plot(),
explore_categorical(), explore_pairs(),
ridgeline_plot() and outlier_plot() drop rows
with a missing group, with a message, and give no colour to a level
without rows, so the colours of the levels after it can shift.ridgeline_plot() keeps each ridge’s colour when an
earlier group is too sparse to draw, where the later ridges shifted
colour. group_comparison_plot() and
raincloud_plot() leave a level without rows off the axis,
where they kept an empty place for it with a warning.explore_distribution() and ecdf_plot()
draw a numeric group code, such as blocks numbered 1 and 2, as groups,
where they drew one ungrouped curve.explore_categorical() orders its groups by level, where
it sorted them alphabetically.scatter_trend(),
explore_distribution(), ecdf_plot(),
raincloud_plot(), group_comparison_plot(),
explore_categorical(), ridgeline_plot() and
outlier_plot() warn once past eight groups, when the plot
is made, where they warned each time the plot was drawn.timeseries_plot() names the
average of an even rolling window as, for example, ‘2×12
centred moving average’, where it said ‘12-point’.timeseries_plot() keeps the level order of a factor
group, where it ordered the groups by first appearance, and
leaves out unused levels, so the legend order and colours can
change.decompose_plot(method = "classical") leaves out missing
values at either end of a series, so its panels start at the first
observation and end at the last, where they spanned the missing
stretch.explore_bivariate() treats dates, date-times and
durations as numeric, where it took them for categories, so they give a
scatter plot with a trend or box plots by level.threshold_plot() draws its metrics as steps that change
at each score, where it joined adjacent scores with straight lines.dumbbell_plot(sort = "none") keeps a factor’s level
order, or else the order in which the categories first appear, where it
sorted them alphabetically.survival_plot() drops an arm whose every
status is missing, also when group is a
factor, where it kept it in the legend with no curve and made
logrank = TRUE fail.text_contrast row of check_figure()
reports the text furthest short of its own threshold, with that
threshold and the text’s printed size and weight, and a translucent
colour is named with its alpha.check_figure() say when colour is a
continuous scale, which they do not audit, where they said that nothing
was distinguished by colour.check_figure() say whether
their closest pair are colours or fill colours, and its
text_size and text_contrast rows no longer say
that a figure draws no text when all of its text is inside the
panel.pca_plot() and
silhouette_plot() draw their labels without a border or a
warning about linewidth, as on later versions.citation("depictr") gives the DOI that CRAN registered,
https://doi.org/10.32614/CRAN.package.depictr, where it
gave the Zenodo concept DOI, https://doi.org/10.5281/zenodo.21266308.simulate_figure() shows a finished ggplot or patchwork,
legend and colourbar included, as a reader with a colour-vision
deficiency sees it, and returns a gtable to draw or save.depictr_palette(order = "contrast"), also in
scale_colour_depictr() and
scale_fill_depictr(), reorders the palette so that up to
three groups pass every colour check of check_figure() and
four keep 3:1 against white. depictr_options(palette = )
can make it the default.exponentiate = TRUE in coefficient_plot()
and compare_models() gives odds, rate or risk ratios, which
the plots draw on a log axis with the reference line at 1 and an axis
title naming the ratio.check_figure() audits a patchwork composite, the
package’s own included, panel by panel, with a leading
panel column, where it refused one.check_figure(max_text_pt = ) checks a ceiling on text
size in a text_size_max row, and
save_plot(check = TRUE) audits a figure at the width it is
saved at, gives one warning that names each failed check and attaches
the report as attr(, "check").power_curve_plot(solved = ) marks a solved sample size,
such as the result of pilotr’s target_n(), with a dashed
line and its interval as a grey band.power_curve_plot() reads the output of pilotr’s
power_curve_mixed() as it is, with its
n_subject column and its band, and gains n,
power, group, lower,
upper and y_lab after its existing arguments.
Several curves are told apart by colour and line type.coefficient_plot(standardise = TRUE, sd_multiple = 2)
scales continuous inputs by two standard deviations, Gelman’s (2008)
convention.ts_forecast() and timeseries_plot() gain
lambda, for a Box-Cox or log transformation, and
decompose_plot() gains s_window, the STL
seasonal window.posterior_plot() and
frequentist_bayesian_plot() format each parameter’s label
once, and format_terms() is much faster on long vectors, so
a draws object with many draws no longer takes minutes to plot.pca_plot() takes an unquoted group column
name and refuses a prcomp fit that holds no scores, with a
message that says so.vif_plot() refuses a model with aliased coefficients
and names their terms, where it failed with an unrelated base-R
error.correlation_heatmap() refuses a column in
cols that it cannot correlate, such as a factor or a date,
and names it, where it failed with an error that did not.explore_pairs() refuses a numeric group
named in cols with a message that says so, where it failed
with an error from cor().decompose_plot() and ts_forecast() refuse
a missing value inside the series with a message that names
x, and a series of exactly two cycles gets the package’s
own message.estimation_plot() and
group_comparison_plot(differences = TRUE) work in a fresh R
session, where they failed unless patchwork had been loaded.effects_plot(), interaction_plot() and
model_report() work with a transformed term, such as
log(), poly() or ns(), and with
an offset, held at its mean on the link scale, where they failed. A name
such as "log(rainfall)" stands for the variable it
transforms.random_effects_plot() draws models with several
uncorrelated or split random-effect terms per grouping factor, such as
(1 + x || g), fitted with lmer() or
glmer(), where it failed.glm fitted with na.action = na.exclude
works in the ROC, precision-recall, threshold, gain and lift plots and
in confusion_matrix_plot(), which failed.posterior_plot() reads a posterior draws
object made from a brms fit, on which it failed with an error about
duplicated factor levels.posterior_plot() accepts an unnamed labels
vector with one label per parameter, which it refused.frequentist_bayesian_plot() with summaries no longer
warns about labels keys it was not given.ts_forecast(), the forecast overlay of
timeseries_plot() and decompose_plot() with
its default STL method read a one-column ts and trim
missing values at either end of a series, where the STL decomposition
failed. A forecast starts after the last observation.seasonal_plot() accepts a frequency that is not a whole
number, such as 365.25 / 7, and rounds it down to whole seasons with a
message, where it refused every such frequency.scatter_trend() accepts a numeric group code, which
failed with a discrete-scale error.explore_pairs() leaves a numeric group out
of its default columns, where it failed with an error from
cor().coefficient_plot(),
frequentist_bayesian_plot() and
posterior_plot() no longer warn about an unknown
inherit.aes.metric = "cie76" in palette_safety() and
check_figure(), kept to reproduce colour distances computed
before this release, gives a warning once per session. The default,
"ciede2000", replaces it, and a later release will remove
it.depictr_palette(),
depictr_options() and palette_safety() no
longer call the built-in palette the Okabe-Ito set. The help page of
depictr_palette() says that it keeps six of that set’s
eight colours, with the brand blue and a grey in place of its blue and
black.explore_pairs() refuses columns beyond
max_cols, posterior_plot() refuses an unknown
style and depictr_palette(n = NULL) gives
seven colours for a ramp. depictr_options() and
theme_depictr() no longer present
depictr_brand(), depictr_accent() and
depictr_reference() as functions to call, since the package
does not export them. cluster_plot(seed = NULL) draws from
the session’s random-number stream, coefficient_plot()
applies interaction to the terms a named
labels vector leaves out and survival_plot()
tests a survfit object with the base-R log-rank test.
missingness_map() takes its missing-cell colour from the
accent, k_diagnostic() takes the elbow farthest from the
chord, dumbbell_plot() puts the shortest dumbbell at the
top only when every difference is negative and
estimation_plot() no longer says that zero lines up with
the reference group’s mean.vignette("exploring-data") says that
summary_table() takes its column names as strings and that
an argument naming several columns, such as cols, takes a
character vector, where it said that column names could be given quoted
or unquoted. It also says that, of the model-estimate plots, only
coefficient_plot() takes standardise and that
every plotting function returns a ggplot, where it said that all of them
took standardise and that every function returned a ggplot.
vignette("model-estimates") and the help page of
coefficient_plot() no longer claim that standardised
magnitudes are comparable. The help page of
threshold_plot() and
vignette("diagnostics-and-uncertainty") say that the
maximum-F1 threshold is drawn dotted and the Youden threshold dashed,
where they said both were dashed.
vignette("multivariate-and-survival") says that only an arm
whose curve reaches 0.5 gets a median guide, and
vignette("time-series") says that STL holds the seasonal
component fixed under the default s_window = "periodic",
where it said that STL lets it change from year to year.check_figure() gains a section on
auditing a figure at the width it is printed at, with the text floors of
four publishers, where it called 6 pt a common publisher floor, and says
that four colours of the default palette fall below 3:1 against white.
The help page of save_plot() describes the formats that
three publishers accept.vignette("depictr") shows a figure drawn by
simulate_figure() and covers four publishers’ text floors,
Nature’s ceiling, the contrast-first palette order and how the audit
treats a continuous colour scale.check_figure() and
vignette("depictr") cite Crameri et al. (2020) for
preferring a perceptually uniform map, such as viridis, to a rainbow,
and no longer for depictr’s own ramps. The help page of
depictr_palette() no longer calls the Okabe-Ito set ‘widely
recommended’, and that of model_fit_table() credits Cameron
and Windmeijer (1997) with the Kullback-Leibler R-squared for
exponential-family models only. The help page of
save_plot() capitalises the word after the colon in the
title of Nature’s figure guide, and that of k_diagnostic()
gives the year of its citation of Tibshirani et al. (2001).check_figure(), an accessibility and honesty audit
of the figure you are about to submit. Until now the package could vouch
for its palette and say nothing about a finished plot, which is the
thing a reader sees. Give it anything a depictr function returns,
including a plot extended afterwards with +, and it
introspects the build and returns a tidy table. The rows cover the
separability of the encoding colours under each dichromacy and in
greyscale, the smallest text size against a stated physical output
width, the WCAG contrast of the text and of the geometry against their
backgrounds, and whether any distinction is carried by colour alone.
Every row carries the value it measured beside the threshold it was
measured against, so a verdict can be argued with.simulate_cvd()
and palette_safety() were internal here while the Python
twin exported both, so the advertised parity did not hold. They now
match in name, arguments, return shape and refusal wording.
simulate_cvd() gains a severity argument and
returns lower-case hex, as the Python twin does.
palette_safety() returns the full report, naming the worst
condition, the closest pair and the verdict. The old return was a bare
vector of distances.#e69f00) and sky blue
(#56b4e9) differ by 0.79 in CIE lightness, so a
black-and-white printer renders them as the same grey. The Okabe-Ito
guarantee is about hue confusion and was never a claim about greyscale.
The threshold stays where it is, and vignette("depictr")
now states the limitation where the claim is made, so the package’s own
defaults are held to the same standard as anybody else’s.quantile_residuals() produced nonsense for a
cbind(successes, failures) binomial model. The two-column
response matrix was flattened to a vector and the raw success counts
were then multiplied by the trial totals a second time, so the residuals
of a well-specified model centred far from zero and the Q-Q diagnostic
looked catastrophically misspecified. The matrix response now supplies
its counts and trial totals directly, and the residuals are standard
normal again where they should be.ridgeline_plot() stacked its overlaps upside down. The
row sort meant to draw the top ridge first is a no-op, because ggplot2
draws ribbon groups in factor-level order, so each upper ridge painted
over the one below it, the opposite of the conventional ridgeline
overlap. The draw order is now carried by the group aesthetic, with the
colour assignment unchanged.random_effects_plot(sort = TRUE) froze every facet in
the first facet’s order. With more than one term the level factor was
shared across panels, so only the first panel came out sorted and the
rest zigzagged. Each facet now orders its own levels, with the plain
level names kept on the axis.title in
power_curve_plot() is no longer run through
format_terms(), which turned a colon into a multiplication
sign and blanked underscores. Only a title recovered from the
power-curve object, which is a raw term name, is tidied.survival_plot() drew a phantom arm for a group that
does not exist. An NA in group became a level
of its own, matching no observation, so the plot gained an all-censored
curve for a group nobody was in, and under logrank = TRUE
it failed outright. Missing groups are now dropped with a message saying
how many.survival_plot() silently discarded non-finite follow-up
times. Dropping a case from a Kaplan-Meier fit changes the denominator,
and so every step of the curve and every cell of the number-at-risk
table, while the figure carries no trace of it. It now refuses them,
with the same message as the Python twin, since whether to drop or
impute is the analyst’s decision to make. This replaces a silent drop,
so it is a behaviour change for anyone who relied on the old
handling.depictr_palette() interpolated past its accessibility
guarantee without saying so. Beyond the eight Okabe-Ito base colours the
palette is a ramp, and the colour-vision-deficiency guarantee that is
this package’s reason for existing stops holding, so the interpolated
palette fails the package’s own safety check. It now warns at the point
of interpolation, and only for the built-in palette, since a
user-supplied one carries no such claim. The documentation is qualified
to match.summary_table() counted missing-group records in
Overall and in no group column, so the per-group sizes
silently fell short of the headline N. They now get a
Missing column of their own.compare_models(facet = TRUE) now honours
depictr_options(reference = ) for its per-panel reference
line. seasonal_plot() no longer labels a frequency-7 series
Mon..Sun, an alignment a plain ts cannot know. And a
mistyped labels key now warns, so the raw parameter name no
longer sits on the plot.survival_plot() now says when it drops observations
with a missing status, in the wording of the missing-group message. The
drop was previously silent. A status that is missing for every
observation is an error. This brings the third kind of incomplete
survival input into line with the other two, a missing group being
announced and a non-finite time refused.silhouette_plot() checks that clusters has
one entry per row of data before dropping incomplete rows.
A vector sized to the complete rows used to slip past the late check,
because subsetting it with the logical index padded it with
NA, and then died in the distance computation. It is now
refused with the same message cluster_plot() uses.tidy_estimates() no longer fails on a rank-deficient
lm with a raw “differing number of rows” error.
confint() keeps aliased terms as NA rows while
coef(summary()) drops them, so the intervals are now cut to
the estimated terms, with a message naming the aliased terms that were
left out.cluster_plot() and k_diagnostic() drop
zero-variance columns with a message when scale = TRUE, as
correlation_heatmap() already did, so a raw k-means error
no longer escapes. k_diagnostic() now names the values of
k_range it cannot evaluate instead of dropping them from
the search without a word, and explore_pairs() labels an
undefined correlation n/a where it once printed
r = NA beside a raw stats::cor() warning.model_report() at all, because
patchwork::free(type =, side =) arrived in 1.3.0.gain_plot() documents that the perfect-model reference
line is drawn for a single model only, which is what the code has always
done, and a test now pins it. The line bends at the prevalence of the
outcome, and overlaid models need not share a prevalence.model_fit_table() documents that a single model is
enough, which is what the code always accepted, and
raincloud_plot() no longer claims to be built from base
graphics primitives: like ridgeline_plot(), it is base R
and ggplot2 alone.?depictr again lists every exported function
(scale_fill_depictr() and the
scale_color_depictr() alias were missing), the README no
longer describes the Python package as a feature-parity twin (its own
README says otherwise), CONTRIBUTING no longer claims the maintenance
workflows close their own issues, and the
standardise = TRUE axis label now names the x-only
convention the figure uses.vif_plot() restricts its scale to the severity levels
present, which removes an empty key entry from the rendered figure.depictr_palette()
returning the palette’s hex colours directly, ready for
scale_fill_manual() or a base-graphics col =
argument, and shows depictr_options() setting defaults for
every later plot and returning the previous values, so the earlier look
can be put back afterwards.vignette("diagnostics-and-uncertainty") saves an
arranged panel with save_plot(), which writes at a
print-ready 300 dpi by default and creates any missing directories.vignette("multivariate-and-survival") adds
k_diagnostic(method = "gap"), which compares within-cluster
dispersion against a null reference and so, unlike the other two
criteria, can support k = 1.vignette("time-series") adds classical decomposition,
which holds the seasonal component fixed across the series where STL
lets it evolve from year to year.optimizer_fixef_plot() and
power_curve_plot() reference pages describe what each plot
shows. Both pages previously described the prototype gists the plots
grew from.roc_curve_plot() rejects a ci that
resolves to fewer than one bootstrap resample, with a clear error. Until
then it drew an all-NA band and an [NA, NA]
AUC annotation, saying nothing.wellbeing_survey, region now shifts stress and
income, which flow through to life satisfaction. The region-grouped
plots (faceted densities, ridgelines, the region dendrogram) therefore
compare four distinct distributions where earlier they compared four
samples of one. The bundled datasets are regenerated by
data-raw/generate_datasets.R as before.inst/CITATION and
CITATION.cff) carries the Zenodo concept DOI, and the
citation title uses sentence case.?depictr) lists every exported
function and all five bundled datasets. Previously several functions and
two datasets were missing.vif_plot() example fits deliberately collinear
predictors, so the plot shows inflated VIFs sitting above the threshold
line. The earlier example produced near-identical bars around 1, with
the line off the axis.depictr_options() describes what brand and
accent actually drive, and depictr_palette()
notes that qualitative colours interpolated beyond the base set are not
guaranteed to stay distinguishable under colour-vision deficiency.lift_plot() documents its own top-right inside-legend
corner rather than inheriting gain_plot()‘s bottom-right
wording, and DESCRIPTION notes that composite panels return
’patchwork’ objects.LICENSE.md carries the full MIT text
so the site’s licence page renders in full, as in the sibling
packages.posterior_plot(),
residual_diagnostics_plot(), decompose_plot())
are wrapped in \donttest{} so each example file stays
within CRAN’s five-second budget. The first call of every example still
runs.CITATION.cff and the test artefact
Rplots.pdf are excluded from the build tarball, and the
test that produced Rplots.pdf draws to a null device
instead.legend_inside gates. The legend moves
inside the panel when a plot’s gate is satisfied, and the theme is left
alone when it is not.LICENSE file, and added
community and contribution files.First release. depictr is a unified, consistent toolkit of
publication-ready plots spanning the whole analysis workflow. It grew
out of, and generalises, three earlier plotting functions
(frequentist_bayesian_plot, plot.fixef.allFit
and powercurvePlot).
explore_distribution(),
explore_categorical(), explore_bivariate(),
explore_pairs(), correlation_heatmap(),
missingness_map(), outlier_plot(),
raincloud_plot(), group_comparison_plot(),
scatter_trend() and summary_table().estimation_plot() for estimation statistics: group
effect sizes (mean differences, Cohen’s d / Hedges’ g)
with bootstrap confidence intervals, in the spirit of the ‘new
statistics’.ecdf_plot() (empirical cumulative distribution,
optionally by group), ridgeline_plot() (overlapping
per-group densities) and dumbbell_plot() (a connected
two-group comparison across categories).explore_distribution() gains facet to draw
one panel per group instead of overlaying them (much clearer beyond a
few groups), and correlation_heatmap() gains
reorder to cluster correlated variables together.pca_plot() and scree_plot() (principal
component analysis), cluster_plot() (k-means on
principal-component axes) and dendrogram_plot()
(hierarchical clustering), and survival_plot()
(Kaplan-Meier curves with a number-at-risk table, median survival and an
optional log-rank test, all computed in base R).silhouette_plot() and k_diagnostic() help
choose and validate the number of clusters (silhouette widths, plus
elbow and average-silhouette diagnostics).timeseries_plot() (one or more series with an optional
moving average), acf_plot() (autocorrelation / partial
autocorrelation) and decompose_plot() (trend / seasonal /
remainder decomposition).seasonal_plot() (seasonal subseries) and
ts_forecast() (a simple, dependency-free forecast with
prediction intervals).tidy_estimates() provides the shared tidy estimate
table, with methods for lm, glm,
merMod and data frames and a fallback to
broom::tidy().coefficient_plot(), compare_models(),
frequentist_bayesian_plot(), effects_plot(),
interaction_plot(), random_effects_plot(),
optimizer_fixef_plot() and
model_fit_table().frequentist_bayesian_plot() now draws the full Bayesian
posterior for each term as a half-eye density and overlays the matching
frequentist point estimate and confidence interval, so the two
inferential frameworks can be compared directly. It reads posterior
draws from brmsfit, stanreg,
draws/matrix objects or a data frame.residual_diagnostics_plot(),
influence_plot(), qq_plot(),
vif_plot(), roc_curve_plot(),
pr_curve_plot(), gain_plot(),
lift_plot(), calibration_plot() and
confusion_matrix_plot().binned_residual_plot() (binned residuals for logistic
and other GLMs, with approximate error bounds) and
threshold_plot() (classification metrics across decision
thresholds, highlighting Youden’s J and the maximum-F1
cut-off).posterior_plot() summarises posterior draws with a
choice of styles ("halfeye", "interval",
"gradient" or "dots") and can annotate a
region of practical equivalence (ROPE) and the probability of
direction.power_curve_plot() for power-analysis curves (e.g. from
simr).theme_depictr(), depictr_palette(),
scale_colour_depictr() (and
scale_color_depictr(), scale_fill_depictr()),
palette_preview(), format_terms(),
model_report() (a one-figure model overview),
arrange_plots() and save_plot().depictr_options() sets package-wide defaults once,
covering the brand and accent colours, qualitative palette, base font
size and family, and the colour used for missing values. Every plot and
scale then honours them.coefficient_plot(), compare_models(),
posterior_plot() and
frequentist_bayesian_plot() gain a
facet/scales option that lays each term out in
its own free-scaled panel, so terms on very different scales (a large
intercept alongside small slopes) stay legible instead of being squished
onto the zero line. frequentist_bayesian_plot() uses this
layout by default.silhouette_plot() cluster labels no longer clip.
raincloud_plot() uses one colour per group across all
layers. dendrogram_plot() hides leaf labels for large
trees. confusion_matrix_plot() picks each label’s colour
from the tile luminance. gain_plot() and
lift_plot() label their reference lines.
timeseries_plot() shows a single legend, and
k_diagnostic() now returns the diagnostic curve as a
plot.coefficient_plot() gains standardise,
scaling each coefficient by its predictor’s standard deviation so
magnitudes are comparable. model_report() uses it by
default, removing the empty band in its coefficient panel.vif_plot() shows the ordinary VIF (not its square root)
for single-df terms, scales the axis to the data, and draws a single
clearly-labelled threshold line (reported in the caption when it is
off-axis), leaving no wide empty band or hard-to-read guides.seasonal_plot(style = "season") reverses its sequential
legend so the darkest, most-recent cycle sits at the top, matching the
plotted order.coefficient_plot(),
compare_models() and
frequentist_bayesian_plot(), where they are read from the
model, so conditionunrelated becomes condition
and word_frequency becomes word frequency.
optimizer_fixef_plot() and posterior_plot()
gain a labels argument for the same. Any user-supplied
labels take precedence. pca_plot() likewise
shows underscores in its loading-arrow labels as spaces
(soil_ph -> soil ph).silhouette_plot() (the bands are labelled in place) and
cluster_plot() when the centroids are labelled.survival_plot() has been tidied in several ways. The
log-rank annotation renders a proper chi-squared and an italic
p, formatted APA style (no leading zero, p < .001
below that threshold). The median guide is labelled
median <value>. The y-axis title margin is tighter,
and the colour legend and the number-at-risk table list the groups in
the same order, following the group factor’s levels.legend_inside argument (off by default) draws the
legend inside the panel, over a semi-transparent background, in a corner
the plot usually leaves empty, which reclaims the right-hand margin. It
is offered by roc_curve_plot(), gain_plot(),
lift_plot() (bottom-right / top-right of the curve),
ecdf_plot(), survival_plot(),
explore_distribution(), dumbbell_plot() and
missingness_map(). For any other plot the same is one
theme() call, and vignette("exploring-data")
shows how, alongside tidying legend titles.theme_depictr() now centres legend titles over their
keys, which reads more tidily than ggplot2’s default left alignment,
especially for an inside or a top/bottom legend.estimation_plot() reserves more headroom above the
lower panel so the effect-size annotation (Hedges’ g / Cohen’s
d) is never clipped.scree_plot() colour-matches and names its dual axes,
‘Variance explained (bars)’ on the left and ‘Cumulative (line)’ on the
right.model_report()’s n and R, and
estimation_plot()’s g / d.crop_yield
column is now fertiliser,
coefficient_plot()/model_report() take
standardise, and confusion_matrix_plot() takes
normalise.lexical_decision
(counterbalanced priming experiment), wellbeing_survey
(with realistic missingness), crop_yield (a
fertiliser-by-treatment field trial), clinical_trial
(right-censored survival with a rare adverse event) and
monthly_sales (two seasonal retail series).depictr_palette()
provides sequential and diverging variants.
palette_preview() can show any one, or all three, and can
simulate deuteranopia, protanopia or tritanopia so a palette’s
legibility can be checked directly.lme4, broom,
simr, survival, brms,
posterior, ggdist, cluster,
boot) are in Suggests and used only when
available, so the package installs and checks without them. Vignettes
draw on small precomputed model fits shipped in
inst/extdata/, so they knit without a Bayesian or
mixed-model toolchain.seed
(cluster_plot(), qq_plot() and
residual_diagnostics_plot()) restore the caller’s random
number generator state afterwards, so passing one for reproducibility
has no side effect on your own subsequent random draws.