These are adaptations of the R dotchart function that sorts categories
top to bottom, adds auxdata and auxtitle arguments to put
extra information in the right margin, and for dotchart3 adds
arguments cex.labels, cex.group.labels, and
groupfont. By default, group headings are in a larger, bold
font. dotchart3 also cuts a bit of white space from the top and
bottom of the chart. The most significant change, however, is in how
x is interpreted. Columns of x no longer provide an
alternate way to define groups. Instead, they define superpositioned
values. This is useful for showing three quartiles, for example. Going
along with this change, for dotchart3 pch can now be a
vector specifying symbols to use going across columns of x.
x was changed in this way because to put multiple points on a
line (e.g., quartiles) and keeping track of par() parameters when
dotchart2 was called with add=TRUE was cumbersome.
dotchart3 changes the margins to account for horizontal labels.
dotchartp is a version of dotchart3 for making the chart
with the plotly package.
summaryD creates aggregate data using summarize and
calls dotchart3 with suitable arguments to summarize data by
major and minor categories. If options(grType='plotly') is in
effect and the plotly package is installed, summaryD uses
dotchartp instead of dotchart3.
summaryDp is a streamlined summaryD-like function that
uses the dotchartpl function to render a plotly graphic.
It is used to compute summary statistics stratified separately by a
series of variables.
dotchart3(x, labels = NULL, groups = NULL, gdata = NULL,
cex = par("cex"), pch = 21, gpch = pch, bg = par("bg"),
color = par("fg"), gcolor = par("fg"), lcolor = "gray",
xlim = range(c(x, gdata), na.rm=TRUE), main = NULL, xlab = NULL,
ylab = NULL, auxdata = NULL, auxtitle = NULL, auxgdata=NULL,
axisat=NULL, axislabels=NULL,
cex.labels = cex, cex.group.labels = cex.labels * 1.25,
cex.auxdata=cex, groupfont = 2,
auxwhere=NULL, height=NULL, width=NULL, ...)dotchartp(x, labels = NULL, groups = NULL, gdata = NULL,
xlim = range(c(x, gdata), na.rm=TRUE), main=NULL,
xlab = NULL, ylab = '', auxdata=NULL, auxtitle=NULL,
auxgdata=NULL, auxwhere=c('right', 'hover'),
symbol='circle', col=colorspace::rainbow_hcl,
legendgroup=NULL,
axisat=NULL, axislabels=NULL, sort=TRUE, digits=4, dec=NULL,
height=NULL, width=700, layoutattr=FALSE, showlegend=TRUE, ...)
summaryD(formula, data=NULL, fun=mean, funm=fun,
groupsummary=TRUE, auxvar=NULL, auxtitle='',
auxwhere=c('hover', 'right'),
vals=length(auxvar) > 0, fmtvals=format,
symbol=if(use.plotly) 'circle' else 21,
col=if(use.plotly) colorspace::rainbow_hcl else 1:10,
legendgroup=NULL,
cex.auxdata=.7, xlab=v[1], ylab=NULL,
gridevery=NULL, gridcol=gray(.95), sort=TRUE, ...)
summaryDp(formula,
fun=function(x) c(Mean=mean(x, na.rm=TRUE),
N=sum(! is.na(x))),
overall=TRUE, xlim=NULL, xlab=NULL,
data=NULL, subset=NULL, na.action=na.retain,
ncharsmax=c(50, 30),
digits=4, ...)
the function returns invisibly
a numeric vector or matrix
labels for categories corresponding to rows of
x. If not specified these are taken from row names of x.
see dotchart
a vector of information to be put in the right margin,
in the same order as x. May be numeric, character, or a
vector of expressions containing plotmath markup. For
dotchartp, auxdata may be a matrix to go along
with the numeric x-axis variable, to result in point-specific
hover text.
a column heading for auxdata
similar to auxdata but corresponding to the
gdata argument. These usually represent overall sample sizes
for each group of lines.
a vector of tick mark locations to pass to axis.
Useful if transforming the data axis
a vector of strings specifying axis tick mark labels. Useful if transforming the data axis
number of significant digits for formatting numeric data
in hover text for dotchartp and summaryDp
for dotchartp only, overrides digits to
specify the argument to round() for rounding values for
hover labels
cex for labels
cex for group labels
cex for auxdata
font number for group headings
for summaryD and dotchartp specifies
whether auxdata
and auxgdata are to be placed on the far right of the
chart, or should appear as pop-up tooltips when hovering the
mouse over the ordinary x data points on the chart.
Ignored for dotchart3.
other arguments passed to some of the graphics functions,
or to dotchart3 or dotchartp from summaryD.
The auxwhere='hover' option is a useful argument to pass
from summaryD to dotchartp. Also used to pass
other arguments to dotchartpl from summaryDp.
set to TRUE to put plotly::layout
information in a list as an attribute layout of the
returned plotly object instead of running the
plotly object through the layout function. This
is useful if running dotchartp multiple times to later
put together using plotly::subplot and only then running
the result through plotly::layout.
set to FALSE to suppress the plotly
legend with dotchartp
a formula with one variable on the left hand side (the
variable to compute summary statistics on), and one or two
variables on the right hand side. If there are two variables,
the first is taken as the major grouping variable. If the left
hand side variable is a matrix it has to be a legal R variable
name, not an expression, and fun needs to be able to
process a matrix. For summaryDp there may be more than
two right-hand-side variables.
a data frame or list used to find the variables in
formula. If omitted, the parent environment is used.
a summarization function creating a single number from a
vector. Default is the mean. For summaryDp, fun
produces a named vector of summary statistics, with the default
computing the Mean and N (number of non-missing values).
applies if there are two right hand variables and
groupsummary=TRUE and the marginal summaries over just
the first x variable need to be computed differently
than the summaries that are cross-classified by both
variables. funm defaults to fun and should
have the same structure as fun.
By default, when there are two right-hand
variables, summarize(..., fun) is called a second time
without the use of the second variable, to obtain marginal
summaries for the major grouping variable and display the
results as a dot (and optionally in the right margin). Set
groupsummary=FALSE to suppress this information.
when fun returns more than one statistic and the
user names the elements in the returned vector, you can specify
auxvar as a single character string naming one of them.
This will cause the named element to be written in the right
margin, and that element to be deleted when plotting the statistics.
set to TRUE to show data values (dot
locations) in the right margin. Defaults to TRUE if
auxvar is specified.
an optional function to format values before putting
them in the right margin. Default is the format
function.
a scalar or vector of pch values for ordinary
graphics or a character vector or scalar of plotly
symbols. These correspond to columns of x or elements
produced by fun.
a function or vector of colors to assign to multiple points plotted in one line. If a function it will be evaluated with an argument equal to the number of groups/columns.
see plotly documentation; corresponds to
column names/fun output for plotly graphs only
specify a positive number to draw very faint vertical
grid lines every gridevery x-axis units; for
non-plotly charts
color for grid lines; default is very faint gray scale
specify sort=FALSE to plot data in the original
order, from top to bottom on the dot chart. For
dotchartp, set sort to 'descending' to
sort in descending order of the first column of x, or
'ascending' to do the reverse. These do not make sense
if groups is present.
height and width in pixels for dotchartp if
not using plotly defaults. Ignored for
dotchart3. If set to "auto" the height is
computed using Hmisc::plotlyHeightDotchart.
set to FALSE to suppress plotting of
unstratified estimates
an observation subsetting expression
an NA action function
a 2-vector specifying the number of characters after which an html new line character should be placed, respectively for the x-axis label and the stratification variable levels
Frank Harrell
dotchart,dotchart2,summarize,
rlegend
set.seed(135)
maj <- factor(c(rep('North',13),rep('South',13)))
g <- paste('Category',rep(letters[1:13],2))
n <- sample(1:15000, 26, replace=TRUE)
y1 <- runif(26)
y2 <- pmax(0, y1 - runif(26, 0, .1))
dotchart3(cbind(y1,y2), g, groups=maj, auxdata=n, auxtitle='n',
xlab='Y', pch=c(1,17))
## Compare with dotchart function (no superpositioning or auxdata allowed):
## dotchart(y1, g, groups=maj, xlab='Y')
if (FALSE) {
dotchartp(cbind(y1, y2), g, groups=maj, auxdata=n, auxtitle='n',
xlab='Y', gdata=cbind(c(0,.1), c(.23,.44)), auxgdata=c(-1,-2),
symbol=c('circle', 'line-ns-open'))
summaryDp(sbp ~ region + sex + race + cut2(age, g=5), data=mydata)
}
## Put options(grType='plotly') to have the following use dotchartp
## (rlegend will not apply)
## Add argument auxwhere='hover' to summaryD or dotchartp to put
## aux info in hover text instead of right margin
summaryD(y1 ~ maj + g, xlab='Mean')
summaryD(y1 ~ maj + g, groupsummary=FALSE)
summaryD(y1 ~ g, fmtvals=function(x) sprintf('%4.2f', x))
Y <- cbind(y1, y2) # summaryD cannot handle cbind(...) ~ ...
summaryD(Y ~ maj + g, fun=function(y) y[1,], symbol=c(1,17))
rlegend(.1, 26, c('y1','y2'), pch=c(1,17))
summaryD(y1 ~ maj, fun=function(y) c(Mean=mean(y), n=length(y)),
auxvar='n', auxtitle='N')
Run the code above in your browser using DataLab