combplotp: Combination Plot

In harrelfe/Hmisc: Harrell Miscellaneous

Combination Plot

Description

Generates a plotly attribute plot given a series of possibly overlapping binary variables

Usage

combplotp(
  formula,
  data = NULL,
  subset,
  na.action = na.retain,
  vnames = c("labels", "names"),
  includenone = FALSE,
  showno = FALSE,
  maxcomb = NULL,
  minfreq = NULL,
  N = NULL,
  pos = function(x) 1 * (tolower(x) %in% c("true", "yes", "y", "positive", "+",
    "present", "1")),
  obsname = "subjects",
  ptsize = 35,
  width = NULL,
  height = NULL,
  ...
)

Arguments

formula	a formula containing all the variables to be cross-tabulated, on the formula's right hand side. There is no left hand side variable. If formula is omitted, then all variables from data are analyzed.
data	input data frame. If none is specified the data are assumed to come from the parent frame.
subset	an optional subsetting expression applied to data
na.action	see lm etc.
vnames	set to "names" to use variable names to label axes instead of variable labels. When using the default labels, any variable not having a label will have its name used instead.
includenone	set to TRUE to include the combination where all conditions are absent
showno	set to TRUE to show a light dot for conditions that are not part of the currently tabulated combination
maxcomb	maximum number of combinations to display
minfreq	if specified, any combination having a frequency less than this will be omitted from the display
N	set to an integer to override the global denominator, instead of using the number of rows in the data
pos	a function of vector returning a logical vector with TRUE values indicating positive
obsname	character string noun describing observations, default is "subjects"
ptsize	point size, defaults to 35
width	width of plotly plot
height	height of plotly plot
...	optional arguments to pass to table

Details

Similar to the UpSetR package, draws a special dot chart sometimes called an attribute plot that depicts all possible combination of the binary variables. By default a positive value, indicating that a certain condition pertains for a subject, is any of logical TRUE, numeric 1, "yes", "y", "positive", "+" or "present" value, and all others are considered negative. The user can override this determination by specifying her own pos function. Case is ignored in the variable values.

The plot uses solid dots arranged in a vertical line to indicate which combination of conditions is being considered. Frequencies of all possible combinations are shown above the dot chart. Marginal frequencies of positive values for the input variables are shown to the left of the dot chart. More information for all three of these component symbols is provided in hover text.

Variables are sorted in descending order of marginal frqeuencies and likewise for combinations of variables.

Value

plotly object

Author(s)

Frank Harrell

Examples

if (requireNamespace("plotly")) {
  g <- function() sample(0:1, n, prob=c(1 - p, p), replace=TRUE)
  set.seed(2); n <- 100; p <- 0.5
  x1 <- g(); label(x1) <- 'A long label for x1 that describes it'
  x2 <- g()
  x3 <- g(); label(x3) <- 'This is<br>a label for x3'
  x4 <- g()
  combplotp(~ x1 + x2 + x3 + x4, showno=TRUE, includenone=TRUE)

  n <- 1500; p <- 0.05
  pain       <- g()
  anxiety    <- g()
  depression <- g()
  soreness   <- g()
  numbness   <- g()
  tiredness  <- g()
  sleepiness <- g()
  combplotp(~ pain + anxiety + depression + soreness + numbness +
            tiredness + sleepiness, showno=TRUE)
}

harrelfe/Hmisc documentation built on Nov. 21, 2024, 3:47 p.m.