The argument x can be either:
- a single number (of sets), between 1 and 7
- a metacommand (character) to draw custom intersection zones
- a list, containing values for the different sets: each component is a set,
and only up to 7 components are processed.
- a dataset of boolean values.
A "zone" is a union of set intersections. There are exactly 2^k
intersections in a Venn diagram, where k is the number of sets. To
highlight an entire set, we need a union of all possible intersections which
form that set.
For example, in a 3 sets diagram, the (overall) first set is composed by four
intersections:
100 for what is in the first set but outside sets 2 and outside set 3
101 for the intersection between sets 1 and 3, outside set 2
110 for the intersection between sets 1 and 2, outside set 3
111 for the intersection between all three sets.
A meta-language can be used to define these intersections, using the values of
1 for what is inside the set, 0 for what is outside
the set, and - when its either inside or outside of the set.
The command "1--" is translated as "display only the first, entire
set" is equivalent with the union of the four intersections
"100 + 101 + 110 + 111".
The parameter snames should have the same length as the number of
sets specified by the parameter x.
When the parameter x is used as a metacommand, the number of sets
is calculated as the number of characters in each intersection of the
metacommand. One such character command is "100 + 101 + 110 + 111"
or "1--", and all intersections have exactly three characters.
It is also possible to use a regular, disjunctive normal form, like
"A", which is equivalent with "Abc + AbC + ABc + ABC".
When x is an expression written in DNF, if a valid R statement
then quoting is not even necessary.
The argument snames establishes names for the different sets, or
in its absence it is taken from LETTERS. When x is a
list or a dataframe, snames is taken from their names. The length
of the snames indicates the total number of sets.
A numerical vector can be supplied with the argument ilabels, when
the argument x is a single number of sets. The vector should match
the increasing order of the binary representation for the set intersections.
This argument can also be logical, and if activated with TRUE it constructs
the intersection labels from their particular combinations of 0s and 1s.
Finally, it can also be specified as ilabels = "counts", for counting the
frequency of appearance of each intersection. When the argument x
is a list, the counts are taken from the number of common values for each
intersection, and when x is a data frame, (comprised of exclusively
boolean values 0 and 1) the counts are taken from the number of similar rows. If
a particular intersection does not have any common values (or no rows), the
count "0" is left blank and not displayed in the diagram.
The argument ellipse differentiates between two types of diagrams
for 4 and 5 sets. The idea is to allow for as much space as possible for each
intersection (also as equal as possible) and that is impossible if preserving
the shape of an ellipse. The default is to create large space for the
intersections, but users who prefer an ellipse might want to set this argument
to TRUE.
Colors to fill the desired zones (or entire sets) can be supplied via the
argument zcolor (the default is "bw" black and white,
which means no colors at all). Users can either chose the predefined color style,
using zcolor = "style", or supply a vector of custom colors for
each zone. If only one custom color is supplied, it will be recycled for all
zones.
When using zcolor = "style", any other additional arguments for
the borders are ignored.
A different set of predefined colors is used, when argument x is a
QCA type object (a truth table, either from a class tt or from a
class qca). If custom colors are provided via zcolor,
it should have a length of 3 colors: the first for the absence of the outcome
(0), the second for the presence of the outcome (1),
and the third for the contradictions (C). Remainders have no
color, by default.
The argument ilcs works only if the intersection labels
(ilabels) have information, and it sets the size of the labels via
a cex argument. In the absence of a specific value from the user,
it's default is set to 0.6 for all Venn diagrams with up to five sets, and it
automatically decreases to 0.5 for six sets and 0.45 for seven sets.
Via ..., users can specify additional parameters, mainly for the
outer borders of the sets, as specified by par(),
and since version 1.9 it is also used to pass additional aesthetics parameters
for the ggplot2 graphics. All of them are fed either to the base function
lines() which is responsible with the borders, or
to the function geom_path() from package
ggplot2.
For up to 3 sets, the shapes can be circular. For more than 3 sets, the shape
cannot be circular: for 4 and 5 sets they can be ellipses, while for more than 5
sets the shapes cannot be continous (they might be monotone, but not continous).
The 7 sets diagram is called "Adelaide" (Ruskey, 2005).
The most challenging diagram is the one with 6 sets, where for many years it was
thought a Venn diagram didn't even exist. All diagrams are symmetric, except for
the one with 6 sets, where some of the sets have different shapes. The diagram
in this package is an adaptation from Mamakani, K., Myrvold W. and F. Ruskey (2011).
The argument border can be used only for custom intersections
and/or unions, it has no effect when x is a list, or a data frame,
or a truth table object.
The argument par is used to define a custom set of parameters when
producing the plot, to ensure a square shape of about 15 cm and eliminate the
outer regions. If deactivated, users can define their own size and shape of the
plot using the system function par(). By default,
the plot is always produced using a size of 1000 points for both horizontal and
vertical, unless the argument ggplot is activated, when the
argument par will have no effect.