---
title: "The `adegraphics` package"
author: "Alice Julien-Laferrière, Aurélie Siberchicot and Stéphane Dray"
date: '`r Sys.Date()`'
output:
html_vignette:
number_sections: yes
toc: yes
bibliography: adegraphics.bib
vignette: >
%\VignetteIndexEntry{The `adegraphics` package}
%\VignetteEngine{knitr::rmarkdown}
%\VignetteEncoding{UTF-8}
\usepackage[utf8]{inputenc}
---
<br>
<hr>
<br>
The `adegraphics` package [@Siberchicot2017] is a complete reimplementation of the
graphical functionalities of the `ade4` package [@Dray2007]. The package
has been initially designed to improve the representation of the outputs
of multivariate analyses performed with `ade4` but as its graphical
functionalities are very general, they can be used for other purposes.
The `adegraphics` package provides a flexible environment to produce,
edit and manipulate graphs. We adopted an *object oriented* approach (a
graph is an object) using `S4` classes and methods and used the
visualization system provided by the `lattice` [@Sarkar2008] and `grid`
[@Murrell2005] packages. In `adegraphics`, graphs are R objects that can
be edited, stored, combined, saved, removed, etc.
Note that we tried to facilitate the handling of `adegraphics` by `ade4`
users. Hence, the name of functions and parameters has been preserved in
many cases. The main changes are listed in the appendix of this vignette
so that it should be quite easy to use `adegraphics`. However, several
new functionalities (graphical parameters, creation and manipulation of
graphical objects, etc.) are now available and detailed in this
vignette.
The *adelist* mailing list can be used to send questions and/or comments
on `adegraphics` (see <https://listes.univ-lyon1.fr/sympa/info/adelist>)
<br>
An overview of object classes
=============================
In `adegraphics`, a user-level function produces a plot that is stored
(and returned) as an object. The class architecture of the objects
created by `adegraphics` functions is described in [Figure 1](#classes).
<br>
<div style="text-align:center">
<a name="classes"></a>
<span style="color:blue">Figure 1: Classes structure and user-level functions</span>
<img src="classes.png" style="width:700px"/>
</div>
<br>
This class management highlights a hierarchy with two parent classes:
- `ADEg` for simple graphs. It contains the display of a single data
set using only one kind of representation (e.g., arrows, points,
lines, etc.)
- `ADEgS` for multiple graphs. It contains a collection of at least
two simple graphs (`ADEg`, `trellis` or `ADEgS`)
The `ADEg` class has five child classes which are also subdivided in
several child classes. Each of these five child classes is dedicated for
a particular graphical data representation:
- `ADEg.S1`: unidimensional graph of a numeric score
- `ADEg.S2`: bidimensional graph of xy coordinates (`matrix` or
`data.frame` object)
- `ADEg.C1`: bidimensional graph of a numeric score (bar chart or
curve)
- `ADEg.T`: heat map-like representation of a data table (`matrix`,
`data.frame`, `dist` or `table` object)
- `ADEg.Tr`: ternary plot of xyz coordinates (`matrix` or `data.frame`
object)
The `ADEg` class and its five child classes are virtual: it is not
allowed to create object belonging to these classes. Users can only
create objects belonging to child classes by calls to user functions (see the [User functions](#user-functions) section).
Simple graph (`ADEg` object)
============================
In the `adegraphics` package, a graph is created by a call to a user
function and stored as an R object. These functions allow to display the
raw data but also the outputs of a multivariate analysis. The following
sections describe the different graphical functions available in the
package.
<a name="user-functions"></a>
User functions
--------------
Several user functions are available to create a simple graph (stored as
an `ADEg` object in R). Each function creates an object of a given class
(see [Figure 1](#classes)). [Table 1](#functionsADEg) lists the different
functions, their corresponding classes and a short description. The
`ade4` users would not be lost: many functions have kept their names in
`adegraphics`. The main changes are listed in [Table 2](#functionsADEgchanged).
<br>
<a name="functionsADEg"></a>
<span style="color:blue">Table 1: Graphical functions available in `adegraphics`</span>
Function Class of the returned object Description
--------- ------------------------------ --------------
`s1d.barchart` `C1.barchart` 1-D plot of a numeric score by bars
`s1d.curve` `C1.curve` 1-D plot of a numeric score linked by curves
`s1d.curves` `C1.curves` 1-D plot of multiple scores linked by curves
`s1d.density` `C1.density` 1-D plot of a numeric score by density curves
`s1d.dotplot` `C1.dotplot` 1-D plot of a numeric score by dots
`s1d.gauss` `C1.gauss` 1-D plot of a numeric score by Gaussian curves
`s1d.hist` `C1.hist` 1-D plot of a numeric score by bars
`s1d.interval` `C1.interval` 1-D plot of the interval between two numeric scores
`s1d.boxplot` `S1.boxplot` 1-D box plot of a numeric score partitioned in classes
`s1d.class` `S1.class` 1-D plot of a numeric score partitioned in classes
`s1d.distri` `S1.distri` 1-D plot of a numeric score by means/tandard deviations computed using an external table of weights
`s1d.label` `S1.label` 1-D plot of a numeric score with labels
`s1d.match` `S1.match` 1-D plot of the matching between two numeric scores
`s.arrow` `S2.arrow` 2-D scatter plot with arrows
`s.class` `S2.class` 2-D scatter plot with a partition in classes
`s.corcircle` `S2.corcircle` Correlation circle
`s.density` `S2.density` 2-D scatter plot with kernel density estimation
`s.distri` `S2.distri` 2-D scatter plot with means/standard deviations computed using an external table of weights
`s.image` `S2.image` 2-D scatter plot with loess estimation of an additional numeric score
`s.label` `S2.label` 2-D scatter plot with labels
`s.logo` `S2.logo` 2-D scatter plot with logos (pixmap objects)
`s.match` `S2.match` 2-D scatter plot of the matching between two sets of coordinates
`s.Spatial` `S2.label` Mapping of a `Spatial*` object
`s.traject` `S2.traject` 2-D scatter plot with trajectories
`s.value` `S2.value` 2-D scatter plot with proportional symbols
`table.image` `T.image` Heat map-like representation with colored cells
`table.value` `T.value` or `T.cont` Heat map-like representation with proportional symbols
`triangle.class` `Tr.class` Ternary plot with a partition in classes
`triangle.label` `Tr.label` Ternary plot with labels
`triangle.match` `Tr.match` Ternary plot of the matching between two sets of coordinates
`triangle.traject` `Tr.match` Ternary plot with trajectories
<br>
<a name="functionsADEgchanged"></a>
<span style="color:blue">Table 2: Changes in functions names between `ade4` and `adegraphics`</span>
Function in `ade4` Equivalence in `adegraphics`
------------------------------------------- ------------------------------
`table.cont`, `table.dist`, `table.value` `table.value` [^1]
`table.paint` `table.image`
`sco.boxplot` `s1d.boxplot`
`sco.class` `s1d.class`
`sco.distri` `s1d.distri`
`sco.gauss` `s1d.gauss`
`sco.label` `s1d.label`
`sco.match` `s1d.match`
`sco.quant` no equivalence
`s.chull` `s.class`[^2]
`s.kde2d` `s.density`
`s.match.class` superposition of `s.match` and `s.class`
`triangle.biplot` `triangle.match`
`triangle.plot` `triangle.label`
`s.multinom` `triangle.multinom`
[^1]: The `table.value` function is now generic and can handle `dist` or `table` objects as arguments.
[^2]: Convex hulls are now drawn by the `s.class` function (argument `chullSize`.)
Arguments
---------
The list of arguments of a function are given by the `args` function.
```{r label=chunk1}
library(ade4)
library(adegraphics)
args(s.label)
```
Some arguments are very general and present in all user functions:
- `plot`: a logical value indicating if the graph should be displayed
- `storeData`: a logical value indicating if the data should be stored
in the returned object. If `FALSE`, only the names of the data are
stored. This allows to reduce the size of the returned object but it
implies that the data should not be modified in the environment to
plot again the graph.
- `add`: a logical value indicating if the graph should be superposed
on the graph already displayed in the current device (it replaces
the argument `add.plot` in `ade4`).
- `pos`: an integer indicating the position of the environment where
the data are stored, relative to the environment where the function
is called. Useful only if `storeData` is `FALSE`.
- `…`: additional graphical parameters (see below)
Some other arguments influence the graphical outputs and they are thus
specific to the type of produced graph. [Figure 2](#gargsVSclass) summarizes some
of these graphical parameters available for the different functions. We
only reported the parameters stored in the `g.args` slot of the returned
object (see the [Parameters in `g.args`](#gargs) section).
<br>
<div style="text-align:center">
<a name="gargsVSclass"></a>
<span style="color:blue">Figure 2: Specific arguments in each object class</span>
```{r label=fig-gargsVSclass, include=TRUE, echo=FALSE, fig.width=7, fig.height=8}
source("gargsVSclass.R")
```
</div>
<br>
<br>
The `ade4` users would note that the names of some arguments have been
modified in `adegraphics`. The [Appendix](#appendix) gives a full list of
these modifications.
<a name="slots"></a>
Slots and Methods
-----------------
A call to a graphical function (see the [User functions](#user-functions) section) returns an `ADEg`
object. Each object is defined by a number of slots and several methods
are associated to this class. Let us consider the `olympic` data set
available in the `ade4` package. A principal component analysis (PCA) is
applied on the `olympic$tab` table that contains the results for 33
participating athletes at the 1988 summer olympic games:
```{r label=chunk2}
data(olympic)
pca1 <- dudi.pca(olympic$tab, scannf = FALSE)
```
The barplot of eigenvalues is then drawn and stored in `g1`:
```{r label=plot1, fig.height=4, fig.width=4}
g1 <- s1d.barchart(pca1$eig, p1d.horizontal = F, ppolygons.col = "white")
```
<br>
The class of the `g1` object is `C1.barchart` which extends the `ADEg`
class:
```{r label=chunk3}
class(g1)
showClass("C1.barchart")
```
<br>
This object contains different slots:
```{r label=chunk4}
slotNames(g1)
```
<br>
These slots are defined for each `ADEg` object and contain different
types of information. The package `adegraphics` uses the capabilities of
the `lattice` package to display a graph (by generating a `trellis`
object). Hence, several slots contain information that will be passed in
the call to the `lattice` functions:
- `data`: a list containing information about the data.
- `trellis.par`: a list of graphical parameters that are directly
passed to the `lattice` functions using the `par.settings` argument
(see the [Parameters in `trellis.par`](#trellispar) section).
- `adeg.par`: a list of graphical parameters defined in `adegraphics`.
The list of parameters can be obtained using the `adegpar` function
(see the [Parameters in `adeg.par`](#adegpar) section).
- `lattice.call`: a list of two elements containing the information
required to create the `trellis` object: `graphictype` (the name of
the `lattice` functions that should be used) and `arguments` (the
list of parameter values required to obtain the `trellis` object).
- `g.args`: a list containing at least the different values of the
graphical arguments described in [Figure 2](#gargsVSclass) (see the [Parameters in `g.args`](#gargs) section).
- `stats`: a list of internal preliminary computations performed to
display the graph.
- `s.misc`: a list of other internal parameters.
- `Call`: an object of class `call` containing the matched call.
<br>
These different slots can be extracted using the `@` operator:
```{r label=chunk5}
g1@data
```
<br>
All these slots are automatically filled during the object creation. The
`trellis.par`, `adeg.par` and `g.args` can also be modified *a
posteriori* using the `update` method (see the [Customizing a graph](#update) section). This allows to
customize graphs after their creation.
We consider the correlation circle that depicts the correlation between
PCA axes and the results for each event:
```{r label=plot2, fig.width=4, fig.height=4}
g2 <- s.corcircle(pca1$co)
```
```{r label=chunk6}
class(g2)
g2@g.args
```
<br>
The argument `fullcircle` can be updated *a posteriori* so that the
original object is modified:
```{r label=plot3, fig.width=4, fig.height=4}
update(g2, fullcircle = FALSE)
g2@g.args
```
Several other methods have been defined for the `ADEg` class allowing to
extract information, modify or combine objects:
- `getcall`, `getlatticecall` and `getstats`: these accessor methods
return respectively the `Call`, the `lattice.call` and the `stats`
slots.
- `getparameters`: this method returns the `trellis.par` and/or the
`adeg.par` slots.
- `show`, `print` and `plot`: these methods display the `ADEg` object
in the current device or in a new one.
- `gettrellis`: this method returns the `ADEg` object as a `trellis`
object. It can then be exploited using the `lattice` and
`latticeExtra` packages.
- `superpose`, `+` and `add.ADEg`: these methods superpose two `ADEg`
graphs. It returns a multiple graph object of class `ADEgS`
(see the [The basic methods for superposition, juxtaposition and insertion](#superpose) section).
- `insert`: this method inserts an `ADEg` graph in an existing one or
in the current device. It returns an `ADEgS` object
(see the [The basic methods for superposition, juxtaposition and insertion](#superpose) section).
- `cbindADEg`, `rbindADEg`: these methods combine several `ADEg`
graphs. It returns an `ADEgS` object
(see the [The basic methods for superposition, juxtaposition and insertion](#superpose) section).
- `update`: this method modifies the graphical parameters after the
`ADEg` creation. It updates the current display and returns the
modified `ADEg` (see the [Customizing a graph](#update) section).
For instance:
```{r label=chunk7}
getcall(g1) ## equivalent to g1@Call
```
A biplot-like graph can be obtained using the `superpose` method. The
result is a multiple graph:
```{r label=plot4, fig.width=4, fig.height=4}
g3 <- s.label(pca1$li)
g4 <- s.arrow(5 * pca1$c1, add = TRUE)
class(g4)
```
In addition, some object classes have specific methods. For instance, a
`zoom` method is available for `ADEg.S1` and `ADEg.S2` classes. For the
`ADEg.S2` class, the method `addhist` (see the [The basic methods for superposition, juxtaposition and insertion](#superpose) section) decorates a 2-D
graph by adding marginal distributions as histograms and density lines
(this method replaces and extends the `s.hist` function of `ade4`).
```{r label=plot5, fig.width=4, fig.height=4}
zoom(g3, zoom = 2, center = c(2, -2))
```
Multiple graph (`ADEgS` object)
===============================
The `adegraphics` package provides class `ADEgS` to manage easily the
combination of several graphs. This class allows to deal with the
superposition, insertion or juxtaposition of several graphs in a single
object. An object of this class is a list containing several graphical
objects and information about their positioning. Different ways to
generate `ADEgS` objects are described below.
Slots and Methods
-----------------
The class `ADEgS` is used to store multiple graphs. Different slots are
associated to this class (use the symbol `@` to extract information):
- `ADEglist`: a list of graphs stored as `trellis`, `ADEg` and/or
`ADEgS` objects.
- `positions`: a matrix containing the positions of the graphs. It has
four columns and as many rows as the number of graphical objects in
the `ADEglist` slot. For each graph (i.e. row), it contains the
coordinates of the bottom-left and top-right corners in `npc` units
(i.e. normalized parent coordinates varying between 0 and 1).
- `add`: a square binary matrix with as many rows and columns as the
number of graphical objects in the `ADEglist` slot. It allows to
manage the superposition of graphs: the value at the i-th row and
j-th column is equal to 1 if the j-th graphical object is superposed
on the i-th. Otherwise, this value is equal to 0.
- `Call`: an object of class `call` containing the matched call.
Several methods have been implemented to obtain information, edit or
modify `ADEgS` objects. Several methods are inspired from the management
of `list` in R:
- `[`, `[[` and `$`: these methods extract one or more elements from
the `ADEgS` object.
- `getpositions`, `getgraphics` and `getcall`: these methods return
the `positions`, the `ADEglist` and the `Call` slots, respectively.
- `names` and `length`: these methods return the names and number of
graphs contained in the object.
- `[[<-` and `names<-`: these methods replace a graph or its name in
an `ADEgS` object (acts on the `ADEglist` slot).
- `show`, `plot` and `print`: these methods display the `ADEgS` object
in the current device or in a new one.
- `superpose` and `+`: these methods superpose two graphs. It returns
a multiple graph object of class `ADEgS`
(see the [The basic methods for superposition, juxtaposition and insertion](#superpose) section).
- `insert`: this method inserts a graph in an existing one or in the
current device. It returns a multiple graph object of class `ADEgS`
(see the [The basic methods for superposition, juxtaposition and insertion](#superpose) section).
- `cbindADEg`, `rbindADEg`: these methods combine several graphs. It
returns an `ADEgS` object
(see the [The basic methods for superposition, juxtaposition and insertion](#superpose) section).
- `update`: this method modifies the names and/or the `positions` of
the graphs contained in an `ADEgS` object. It updates the current
display and returns the modified `ADEgS`.
We will show in the next sections how these methods can be used to deal
with `ADEgS` objects.
Creating an `ADEgS` object by hand
----------------------------------
The `ADEgS` objects can be created by easy manipulation of several
simple graphs. Some methods (e.g., `insert`, `superpose`) can be used to
create a compilation of graphs by hand.
<a name="superpose"></a>
### The basic methods for superposition, juxtaposition and insertion
The functions `superpose`, `+` and `add.ADEg` allow the superposition of
an `ADEg`/`ADEgS` object on an `ADEg`/`ADEgS` object.
The vector `olympic$score` contains the total number of points computed
for each participant. This vector is used to generate a `factor`
partitioning the participants in two groups according to their final
result (more or less than 8000 points):
```{r label=chunk8}
fac.score <- factor(olympic$score < 8000, labels = c("MT8000", "LT8000"))
```
These two groups can be represented on the PCA factorial map using the
`s.class` function:
```{r label=plot6, fig.width=4, fig.height=4}
g5 <- s.class(pca1$li, fac.score, col = c("red", "blue"), chullSize = 1, ellipseSize = 0,
plabels.cex = 2, pbackground.col = "grey85", paxes.draw = TRUE)
```
The graph with the labels (object `g3`) can then be superposed on this
one:
```{r label=plot7, fig.width=4, fig.height=4}
g6 <- superpose(g5, g3, plot = TRUE) ## equivalent to g5 + g3
class(g6)
```
In the case of a superposition, the graphical parameters (e.g.,
background and limits) of the first graph (the one below) are used as a
reference and applied to the second one (the one above). Note that it is
also possible to use the `add = TRUE` argument in the call of a simple
user function (functions described in [Table 1](#functionsADEg)) to perform
a superposition. The graph `g6` can also be obtained by:
```{r label=chunk9, eval=FALSE}
g5
s.label(pca1$li, add = TRUE)
```
The functions `cbindADEg` and `rbindADEg` allows to combine several
graphical objects (`ADEg`, `ADEgS` or `trellis`) by rows or by columns.
The new created `ADEgS` contains the list of the reduced graphs:
```{r label=plot8, fig.width=6, fig.height=6}
rbindADEg(cbindADEg(g2, g3), cbindADEg(g5, g6), plot = TRUE)
```
The function `insert` allows the insertion of a graphical object on
another one (`ADEg` or `ADEgS`). It takes the position of the inserted
graph as an argument:
```{r label=plot9, fig.width=4, fig.height=4}
g7 <- insert(g2, g6, posi = c(0.65, 0.65, 0.95, 0.95))
class(g7)
```
The different methods associated to the `ADEgS` class allow to obtain
information and to modify the multiple graph:
```{r label=chunk10}
length(g7)
names(g7)
names(g7) <- c("chulls", "labels", "cor")
class(g7[1])
class(g7[[1]])
class(g7$chulls)
```
The multiple graph contains three simple graphs. It can be easily
updated. For instance, the size of the inserted graph can be modified:
```{r label=plot10, fig.width=4, fig.height=4}
pos.mat <- getpositions(g7)
pos.mat
pos.mat[3,] <- c(0.1, 0.7, 0.3, 0.9)
update(g7, positions = pos.mat)
```
The graphs themselves can be modified, without affecting the global
structure of the `ADEgS` object. Here, we replace the correlation circle
by the barplot of eigenvalues:
```{r label=plot11, fig.width=4, fig.height=4}
g7[[3]] <- g1
g7
```
The `addhist` method adds univariate marginal distributions around an
`ADEg.S2` and returns an `ADEgS` object:
```{r label=plot12, fig.width=4, fig.height=4}
addhist(g3)
```
More examples are available in the help page by typing
`example(superpose)`, `example(insert)`, `example(add.ADEg)` and
`example(addhist)` in the R session.
### The `ADEgS` function
The `ADEgS` function provides the most elementary and flexible way to
create an `ADEgS` object. The different arguments of the function are:
- `adeglist`: a list of several `trellis`, `ADEg` and/or `ADEgS`
objects.
- `positions`: a matrix with four columns and as many rows as the
number of graphical objects in the `ADEglist` slot. For each
subgraph, i.e. in each row, the coordinates of the top-right and the
bottom-left hand corners are given in `npc` units (i.e., normalized
parent coordinates varying from 0 to 1).
- `layout`: an alternative way to specify the positions of graphs. It
could be a vector of length 2 indicating the number of rows and
columns used to split the device (similar to `mfrow` parameter in
basic graphs). It could also be a matrix specifying the location of
the graphs: each value in this matrix should be 0 or a positive
integer (similar to `layout` function for basic graphs).
- `add`: a square matrix with as many rows and columns as the number
of graphical objects in the `ADEglist` slot. The value at the i-th
row and j-th column is equal to 1 if the j-th graphical object is
superposed to i-th one. Otherwise, this value is equal to 0.
- `plot`: a logical value indicating if the graphs should be
displayed.
When users fill only one argument among `positions`, `layout` and `add`,
the other values are automatically computed to define the `ADEgS`
object.
We illustrate the different possibilities to create objects with the
`ADEgS` function. Simple juxtaposition using a vector as layout:
```{r label=plot13, fig.width=6, fig.height=3}
ADEgS(adeglist = list(g2, g3), layout = c(1, 2))
```
Layout specified as a matrix:
```{r label=plot14, fig.width=5, fig.height=5}
mlay <- matrix(c(1, 1, 0, 1, 1, 0, 0, 0, 2), byrow = T, nrow = 3)
mlay
ADEgS(adeglist = list(g6, g2), layout = mlay)
```
Using the matrix of positions offers a very flexible way to arrange the
different graphs:
```{r label=plot15, fig.width=5, fig.height=5}
mpos <- rbind(c(0, 0.3, 0.7, 1), c(0.5, 0, 1, 0.5))
ADEgS(adeglist = list(g3, g5), positions = mpos)
```
Lastly, superposition can also be specified using the `add` argument:
```{r label=plot16, fig.width=4, fig.height=4}
ADEgS(list(g5, g3), add = matrix(c(0, 1, 0, 0), byrow = TRUE, ncol = 2))
```
More examples are available in the help page by typing `example(ADEgS)`
in the R session.
Automatic collections
---------------------
The package `adegraphics` contains functionalities to create collections
of graphs. These collections are based on a simple graph repeated for
different groups of individuals, variables or axes. The building process
of these collections is quite simple (definition of arguments in the
call of a user function) and leads to the creation of an `ADEgS` object.
<a name="facets"></a>
### Partitioning the data (`facets`)
The `adegraphics` package allows to split up the data by one variable
(`factor`) and to plot the subsets of data together. This possibility of
conditional plot is available for all user functions (except the
`table.*` functions) by setting the `facets` argument. This is directly
inspired by the functionalities offered in the `lattice` and `ggplot2`
packages.
Let us consider the `jv73` data set. The table `jv73$morpho` contains
the measures of 6 variables describing the geomorphology of 92 sites. A
PCA can be performed on this data set:
```{r label=plot17, fig.width=4, fig.height=4}
data(jv73)
pca2 <- dudi.pca(jv73$morpho, scannf = FALSE)
s.label(pca2$li)
```
The sites are located on 12 rivers (`jv73$fac.riv`) and it is possible
to represent the PCA scores for each river using the `facets` argument:
```{r label=plot18, fig.width=7, fig.height=5.2}
g8 <- s.label(pca2$li, facets = jv73$fac.riv)
length(g8)
names(g8)
```
The `ADEgS` returned object contains the 12 plots. It is then possible
to focus on a given river (e.g., the Doubs river) by considering only a
subplot (e.g., type `g8$Doubs`). The `facets` functionality is very
general and available for the majority of `adegraphics` functions. For
instance, with the `s.class` function:
```{r label=plot19, fig.width=7, fig.height=5.2}
s.class(pca2$li, fac = jv73$fac.riv, col = rainbow(12), facets = jv73$fac.riv)
```
### Multiple axes
All 2-D functions (i.e. `s.*`) returning an object inheriting from the
`ADEg.S2` class have the `xax` et `yax` arguments. These arguments allow
to choose which column of the main argument (i.e. `df`) should be
plotted as x and y axes. As in `ade4`, these two arguments can be simple
integers. In `adegraphics`, the user can also specify vectors as `xax`
and/or `yax` arguments to obtain multiple graphs. Here, we represent the
different correlation circles associated to the first four PCA axes of
the olympic data set:
```{r label=plot20, fig.width=6, fig.height=6}
pca1 <- dudi.pca(olympic$tab, scannf = FALSE, nf = 4)
g9 <- s.corcircle(pca1$co, xax = 1:2, yax = 3:4)
length(g9)
names(g9)
g9@positions
```
### Multiple score
All 1-D functions (i.e. `s1d.*`) returning an object inheriting from the
`ADEg.C1` or `ADEg.S1` classes have the `score` argument. Usually, this
argument should be a numeric vector but it is also possible to consider
an object with several columns (`data.frame` or `matrix`). In this case,
an `ADEgS` object is returned in which one graph by column is created.
For instance for the `olympic` data set, we can represent the link
between the global performance (`fac.score`) and the PCA scores on the
first four axes (`pca1$li`):
```{r label=plot21, fig.width=6, fig.height=6}
dim(pca1$li)
s1d.boxplot(pca1$li, fac.score, col = c("red", "blue"),
psub = list(position = "topleft", cex = 2))
```
### Multiple variable
Some user functions (`s1d.density`, `s1d.gauss`, `s1d.boxplot`,
`s1d.class`, `s.class`, `s.image`, `s.traject`, `s.value`,
`triangle.class`) have an argument named `fac` or `z`. This argument can
have several columns (`data.frame` or `matrix`) so that each column is
used to create a separate graph. For instance, we can represent the
distribution of the 6 environmental variables on the PCA factorial map
of the `jv73$tab` data set:
```{r label=plot22, fig.width=6, fig.height=4}
s.value(pca2$li, pca2$tab, symbol = "circle")
```
### Outputs of the `ade4` package
Lastly, we reimplemented all the graphical functions of the `ade4`
package designed to represent the outputs of a multivariate analysis.
The functions `ade4::plot.*`, `ade4::kplot.*`, `ade4::scatter.*` and
`ade4::score.*` return `ADEgS` objects. It is now very easy to represent
or modify these graphical outputs:
```{r label=plot23, fig.width=6, fig.height=6}
data(meaudret)
pca3 <- dudi.pca(meaudret$env, scannf = FALSE)
pca4 <- dudi.pca(meaudret$spe, scale = FALSE, scannf = FALSE)
coi1 <- coinertia(pca3, pca4, scannf = FALSE, nf = 3)
g10 <- plot(coi1)
class(g10)
names(g10)
g10@Call
```
<a name="update"></a>
Customizing a graph
===================
Compared to the `ade4` package, the main advantage of `adegraphics`
concerns the numerous possibilities to customize a graph using several
graphical parameters. These parameters are stored in slots
`trellis.par`, `adeg.par` and `g.args` (see the [Slots and Methods](#slots) section) of an `ADEg`
object. These parameters can be defined during the creation of a graph
or updated *a posteriori* (using the `update` method).
<a name="trellispar"></a>
Parameters in `trellis.par`
---------------------------
The `trellis.par` slot is a list of parameters that are directly
included in the call of functions of the `lattice` package. The name of
parameters and their default value are given by the `trellis.par.get`
function of `lattice`.
```{r label=chunk11}
library(lattice)
sort(names(trellis.par.get()))
```
Hence, modifications of some of these parameters will modify the
graphical display of an `ADEg` object. For instance, margins are defined
using `layout.widths` and `layout.heights` parameters, `clip` parameter
allows to overpass panel boundaries and `axis.line` and `axis.text`
allow to customize lines and text of axes.
```{r label=plot24, fig.width=7, fig.height=3.5}
d <- scale(olympic$tab)
g11 <- table.image(d, plot = FALSE)
g12 <- table.image(d, axis.line = list(col = "blue"), axis.text = list(col = "red"),
plot = FALSE)
ADEgS(c(g11, g12), layout = c(1, 2))
```
<a name="adegpar"></a>
Parameters in `adeg.par`
------------------------
The `adeg.par` slot is a list of graphical parameters specific to the
`adegraphics` package. The name of parameters and their default value
are available using the `adegpar` function which is inspired by the
`par` function of the `graphics` package.
```{r label=chunk12}
names(adegpar())
```
A description of these parameters is available in the help page of the
function (`?adegpar`). Note that each `adeg.par` parameter starts by the
letter ’p’ and its name relates to the type of graphical element
considered (`ptable` is for tables display, `ppoints` for points,
`parrows` for arrows, etc). Each element of this list can contain one or
more sublists. Details on a sublist are obtained using its name either
as a parameter of the `adegpar` function or after the `$` symbol. For
example, if we want to know the different parameters to manage the
display of points:
```{r label=chunk13}
adegpar("ppoints")
adegpar()$ppoints
```
The full list of available parameters is summarized in [Figure 3](#paramVSparam).
<br>
<div style="text-align:center">
<a name="paramVSparam"></a>
<span style="color:blue">Figure 3: Parameters that can be set with the `adegpar` function.</span>
```{r label=fig-paramVSparam, echo=FALSE, fig.width=7, fig.height=7}
source("paramVSparam.R")
```
</div>
<br>
The ordinate represents the different sublists and the abscissa gives
the name of the parameters available in each sublist. Note that some row
names have two keys separated by a dot: the first key indicates the
first level of the sublist, etc. For example `plabels.boxes` is the
sublist `boxes` of the sublist `plabels`. The parameters `border`,`col`,
`alpha`, `lwd`, `lty` and `draw` in `plabels.boxes` allow to control the
aspect of the boxes around labels.
According to the function called, only some of the full list of
`adeg.par` parameters are useful to modify the graphical display. [Figure 4](#paramVSfunction)
indicates which parameters can affect the display of an object
created by a given user function. For example, the background
(`pbackground` parameter) can be modified for all functions whereas the
display of ellipses (`pellipses` parameter) affects only three
functions.
<br>
<div style="text-align:center">
<a name="paramVSfunction"></a>
<span style="color:blue">Figure 4: Effect of `adeg.par` parameters in `adegraphics` functions.</span>
```{r label=fig-paramVSfunction, echo=FALSE, fig.width=7, fig.height=10}
source("paramVSfunction.R")
```
</div>
<br>
### Global assignment
The `adegpar` function allows to modify globally the values of graphical
parameters so that changes will affect all subsequent displays. For
example, we update the size/color of labels and add axes to a plot:
```{r label=plot25, fig.width=6, fig.height=3}
oldadegpar <- adegpar()
adegpar("plabels")
g13 <- s.label(dfxy = pca1$li, plot = FALSE)
adegpar(plabels = list(col = "blue", cex = 1.5), paxes.draw = TRUE)
adegpar("plabels")
g14 <- s.label(dfxy = pca1$li, plot = FALSE)
ADEgS(c(g13, g14), layout = c(1, 2))
```
As the `adegpar` function can accept numerous graphical parameters, it
can be used to define some graphical themes. The next releases of
`adegraphics` will offer functionalities to easily create, edit and
store graphical themes. Here, we reassign the original default
parameters:
```{r label=chunk14}
adegpar(oldadegpar)
```
### Local assignment
A second option is to update the graphical parameters locally so that
the changes will only modify the object created. This is done using the
dots (`...`) argument in the call to a user function. In this case, the
default values of parameters in the global environment are not modified:
```{r label=plot26, fig.width=4, fig.height=4}
adegpar("ppoints")
s.label(dfxy = pca1$li, plabels.cex = 0, ppoints = list(col = c(2, 4, 5), cex = 1.5, pch = 15))
adegpar("ppoints")
```
In the previous example, we can see that parameters can be either
specified using a ’`.`’ separator or a list. For instance, using
`plabels.cex = 0` or `plabels = list(cex = 0)` is strictly equivalent.
Moreover, partial names can be used if there is no ambiguity (such as
`plab.ce = 0` in our example).
<a name="gargs"></a>
Parameters in `g.args`
----------------------
The `g.args` slot is a list of parameters specific to the function used
(and thus to the class of the returned object). Several parameters are
very general and used in all `adegraphics` functions:
- `xlim`, `ylim`: limits of the graph on the x and y axes
- `main`, `sub`: main title and subtitle
- `xlab`, `ylab`: labels of the x and y axes
- `scales`: a list determining how the x and y axes (tick marks dans
labels) are drawn; this is the `scales` parameter of the `xyplot`
function of `lattice`
The `ADEg.S2` objects can also contain spatial information (map stored
as a `Spatial` object or neighborhood stored as a `nb` object):
- `Sp`, `sp.layout`: objects from the `sp` package to display spatial
objects, `Sp` for maps and `sp.layout` for spatial widgets as a
North arrow, scale, etc.
- `nbobject`: object of class `nb` or `listw` to display neighbor graphs.
When the `facets` (see the [Partitioning the data (`facets`)](#facets) section) argument is used, users can modify the
parameter `samelimits`: if it is `TRUE`, all graphs have the same limits
whereas limits are computed for each subgraph independently when it is
`FALSE`. For example, considering the `jv73` data set, each subgraph is
computed with its own limits and labels are then more scattered:
```{r label=plot27, fig.width=7, fig.height=5.2}
s.label(pca2$li, facets = jv73$fac.riv, samelimits = FALSE)
```
Several other `g.args` parameters can be updated according to the class
of the created object (see [Figure 2](#gargsVSclass)).
Parameters applied on a `ADEgS`
-------------------------------
Users can either apply the changes to all graphs or to update only one
graph. Of an `ADEgS`, to apply changes on all the graphs contained in an
`ADEgS`, the syntax is similar to the one described for an `ADEg`
object. For example, background color can be changed for all graphs in
`g10` using the `pbackground.col` parameter.
```{r label=plot28, fig.width=6, fig.height=6}
g15 <- plot(coi1, pbackground.col = "steelblue")
```
To change the parameters of a given graph, the name of the parameter
must be preceded by the name of the subgraph. This supposes that the
names of subgraphs are known. For example, to modify only two graphs:
```{r label=plot29, fig.width=6, fig.height=6}
names(g15)
plot(coi1, XYmatch.pbackground.col = "steelblue", XYmatch.pgrid.col = "red",
eig.ppolygons.col = "orange")
```
Using `adegraphics` functions in your package
=============================================
In this section, we illustrate how `adegraphics` functionalities can be
used to implement graphical functions in your own package. We created an
objet of class `track` that contains a vector of distance and time.
```{r label=chunk15}
tra1 <- list()
tra1$time <- runif(300)
tra1$distance <- tra1$time * 5 + rnorm(300)
class(tra1) <- "track"
```
For an object of the class `track`, we wish to represent different
components of the data:
- an histogram of distances
- an histogram of speeds (i.e., distance / time)
- a 2D plot representing the distance, the time and the line
corresponding to the linear model that predict distance by time
The corresponding multiple plot can be done using `adegraphics`
functions:
```{r label=plot30, fig.width=7, fig.height=2.3}
g1 <- s1d.hist(tra1$distance, psub.text = "distance", ppolygons.col = "blue",
pgrid.draw = FALSE, plot = FALSE)
g2 <- s1d.hist(tra1$distance / tra1$time, psub.text = "speed", ppolygons.col = "red",
plot = FALSE)
g31 <- s.label(cbind(tra1$time, tra1$distance), paxes = list(aspectratio = "fill",
draw = TRUE), plot = FALSE)
g32 <- xyplot(tra1$distance ~ tra1$time, aspect = g31@adeg.par$paxes$aspectratio,
panel = function(x, y) {panel.lmline(x, y)})
g3 <- superpose(g31, g32)
G <- ADEgS(list(g1, g2, g3))
```
To facilitate the graphical representation of an object of class
`track`, the simplest solution is to design a function `plot` for this
class. We illustrate how to define such function with a particular
emphasis on the management of graphical parameters. The function is
provided below and we detail the different steps.
```{r label=chunk16}
plot.track <- function(x, pos = -1, storeData = TRUE, plot = TRUE, ...) {
## step 1 : sort parameters for each graph
graphsnames <- c("histDist", "histSpeed", "regression")
sortparameters <- sortparamADEgS(..., graphsnames = graphsnames,
nbsubgraphs = c(1, 1, 2))
## step 2 : define default values for graphical parameters
params <- list()
params[[1]] <- list(psub = list(text = "distance"), ppolygons = list(col = "blue"),
pgrid = list(draw = FALSE))
params[[2]] <- list(psub = list(text = "speed"), ppolygons = list(col = "red"),
pgrid = list(draw = FALSE))
params[[3]] <- list()
params[[3]]$l1 <- list(paxes = list(aspectratio = "fill", draw = TRUE))
params[[3]]$l2 <- list()
names(params) <- graphsnames
sortparameters <- modifyList(params, sortparameters, keep.null = TRUE)
## step 3 : create each individual plot (ADEg)
g1 <- do.call("s1d.hist", c(list(score = substitute(x$distance), plot = FALSE,
storeData = storeData, pos = pos - 2), sortparameters[[1]]))
g2 <- do.call("s1d.hist", c(list(score = substitute(x$distance / x$time),
plot = FALSE, storeData = storeData, pos = pos - 2), sortparameters[[2]]))
g31 <- do.call("s.label", c(list(dfxy = substitute(cbind(x$time, x$distance)), plot =
FALSE, storeData = storeData, pos = pos - 2), sortparameters[[3]][[1]]))
g32 <- xyplot(x$distance ~ x$time, aspect = g31@adeg.par$paxes$aspectratio,
panel = function(x, y) {panel.lmline(x, y)})
g3 <- do.call("superpose", list(g31, g32))
g3@Call <- call("superpose", g31@Call, g32$call)
## step 4 : create the multiple plot (ADEgS)
lay <- matrix(1:3, 1, 3)
object <- new(Class = "ADEgS", ADEglist = list(g1, g2, g3), positions =
layout2position(lay), add = matrix(0, ncol = 3, nrow = 3),
Call = match.call())
names(object) <- graphsnames
if(plot)
print(object)
invisible(object)
}
```
In the first step, the arguments given by the user through the dots (…)
argument are managed. A name is given to each subgraph and stored in the
vector `graphnames`. Then, the function `sortparamADEgS` associates the
graphical parameters of the dots (…) argument to each subgraph. If a
prefix is specified and matches the name of a graph (e.g.,
`histDist.pbackground.col = grey`), the parameter is applied only to the
graphic specified (e.g., called `histDist`). If no prefix is specified
(e.g., `pbackground.col = grey`), the parameter is applied to all
subgraphs. The function `sortparamADEgS` returns a list (length equal to
the number of subgraph) of lists of graphical parameters.\
In the second step, default values for some graphical parameters are
modified. The default parameters are stored in a list which has the same
structure that the one produced by `sortparamADEgS` (i.e., names
corresponding to those contained in `graphsnames`). Then, the
`modifyList` function is applied to merge user and defaults values of
paramaters (if a parameter is specified by the user and in the default,
the value given by the user is used).\
In the third step, each subgraph is created. Here, we create two
`C1.hist` objects and superpose a `S2.label` object and a `trellis` one.
The functions `do.call` and `substitute` are used to provide a pretty
call for each subgraph (stored in the `Call` slot).\
In a final step, the multiple graph is build through the creation of a
new `ADEgS` object and possibly plotted.\
The `plot.track` function can then be used by:
```{r label=plot31, fig.width=7, fig.height=2.3}
plot(tra1)
```
Graphical parameters can be modified by:
```{r label=plot32, fig.width=7, fig.height=2.3}
plot(tra1, histDist.ppoly.col = "green", pbackground.col = "grey")
```
Examples
========
Labels customization
--------------------
```{r label=plot33, fig.width=6, fig.height=3}
data(meaudret)
g16 <- s.label(pca3$li, plot = FALSE)
g17 <- s.label(pca3$li, ppoints.col= "red", plabels = list(box = list(draw = FALSE),
optim = TRUE), plot = FALSE)
ADEgS(c(g16, g17), layout = c(1, 2))
```
Ellipses, stars and convex hulls
--------------------------------
```{r label=plot34, fig.width=6, fig.height=6}
g18 <- s.class(pca3$li, fac = meaudret$design$season, plot = FALSE)
g19 <- s.class(pca3$li, fac = meaudret$design$season, ellipseSize = 0,
chullSize = 1, starSize = 0.5, col = TRUE, plot = FALSE)
g20 <- s.class(pca3$li, fac = meaudret$design$season, pellipses.lwd = 2,
pellipses.border = 2:5, pellipses.col = 2:5, plot = FALSE)
g21 <- s.class(pca3$li, fac = meaudret$design$season, ellipseSize = 0,
chullSize = 0, ppolygons.lwd = 2, plines.col = 2:5, starSize = 1.2, plot = FALSE)
ADEgS(c(g18, g19, g20, g21), layout = c(2, 2))
```
Values and legend
-----------------
```{r label=plot35, fig.width=6, fig.height=6}
data(rpjdl)
coa2 <- dudi.coa(rpjdl$fau, scannf = FALSE, nf = 3)
g22 <- s.value(coa2$li, coa2$li[,3], plot = FALSE)
g23 <- s.value(coa2$li, coa2$li[,3], method = "color", ppoints.cex = 0.8,
plegend.size= 0.8, plot = FALSE)
g24 <- s.value(coa2$li, coa2$li[,3], plegend.size = 0.8, ppoints.cex = 0.8,
symbol = "square", method = "color", key = list(columns = 1),
col = colorRampPalette(c("yellow", "blue"))(6), plot = FALSE)
g25 <- s.value(coa2$li, coa2$li[, 3], center = 0, method = "size", ppoints.cex = 0.6,
symbol = "circle", col = c("yellow", "red"), plot = FALSE)
ADEgS(c(g22, g23, g24, g25), layout = c(2, 2))
```
1-D plot
--------
```{r label=plot36, fig.width=6, fig.height=6}
score1 <- c(rnorm(1000, mean = -0.5, sd = 0.5), rnorm(1000, mean = 1))
fac1 <- rep(c("A", "B"), each = 1000)
g26 <- s1d.density(score1, fac1, pback.col = "grey75", plot = FALSE)
g27 <- s1d.density(score1, fac1, col = c(2, 4), plot = FALSE)
g28 <- s1d.density(score1, fac1, col = c(2, 4), p1d.reverse = TRUE, p1d.horizontal = FALSE,
p1d.rug.draw = FALSE, plot = FALSE)
g29 <- s1d.density(score1, fac1, col = c(2, 4), ppolygons.alpha = 0.2,
p1d = list(rug = list(tck = 1, line = FALSE)), plot = FALSE)
ADEgS(c(g26, g27, g28, g29), layout = c(2, 2))
```
Maps and neighbor graphs
------------------------
```{r label=plot37, fig.width=6, fig.height=3}
if(require(Guerry)) {
library(sp)
data(gfrance85)
region.names <- data.frame(gfrance85)[, 5]
col.region <- colors()[c(149, 254, 468, 552, 26)]
g30 <- s.class(coordinates(gfrance85), region.names, porigin.include = FALSE, plot = FALSE)
g31 <- s.class(coordinates(gfrance85), region.names, ellipseSize = 0, starSize = 0,
Sp = gfrance85, pgrid.draw = FALSE, pSp.col = col.region[region.names], pSp.alpha = 0.4,
plot = FALSE)
ADEgS(c(g30, g31), layout = c(1, 2))
}
```
```{r label=plot38, fig.width=6, fig.height=4}
# if(require(Guerry)) {
# s.Spatial(gfrance85[,7:12])
# }
```
```{r label=plot39, fig.width=6, fig.height=3}
data(mafragh, package = "ade4")
g32 <- s.label(mafragh$xy, nb = mafragh$nb, plot = FALSE)
g33 <- s.label(mafragh$xy, nb = mafragh$nb, pnb.ed.col = "red", plab.cex = 0,
pnb.node = list(cex = 3, col = "blue"), ppoints.col = "green", plot = FALSE)
ADEgS(c(g32, g33), layout = c(1, 2))
```
Ternary plots
-------------
```{r label=plot40, fig.width=6, fig.height=3}
data(euro123, package = "ade4")
df <- rbind.data.frame(euro123$in78, euro123$in86, euro123$in97)
row.names(df) <- paste(row.names(euro123$in78), rep(c(1, 2, 3), rep(12, 3)), sep = "")
g34 <- triangle.label(df, label = row.names(df), showposition = TRUE, plot = FALSE)
g35 <- triangle.label(euro123$in78, plabels.cex = 0, ppoints.cex = 2, addmean = TRUE,
show = FALSE, plot = FALSE)
ADEgS(c(g34, g35), layout = c(1, 2))
```
<a name="appendix"></a>
Appendix
=============================
This appendix summarizes the main changes between `ade4` and
`adegraphics`. Each line corresponds to a graphical argument defined in
`ade4` and its equivalent in `adegraphics` is given.
| Arguments in `ade4` | Functions in `ade4` | `g.args` in `adegraphics` | `adeg.par` in `adegraphics` | |
| ------------------- | ----------------------| ----------------------------| ------------------------------|---|
| `abline.x` | `table.cont` | `ablineX` | | |
| `abline.y` | `table.cont` | `ablineY` | | |
| `abmean.x` | `table.cont` | `meanX` | | |
| `abmean.y` | `table.cont` | `meanY` | | |
| `addaxes` | `s.arrow`, `s.chull`, `s.class`, `s.distri`, `s.image`, `s.kde2d`, `s.label`, `s.logo`, `s.match`, `s.traject`, `s.value`, `triangle.class`, `triangle.plot` | |`paxes.draw` | |
| `area` | `s.arrow`, `s.chull`, `s.class`, `s.distri`, `s.image`, `s.kde2d`, `s.label`, `s.logo`, `s.match`, `s.traject`, `s.value` | `Sp` | | a `Sp` object |
| `axesell` | `s.class`, `s.distri`, `triangle.class` | | `pellipses.axes.draw` | |
| `box` | `s.corcircle`, `triangle.plot` | | `pbackground.box` | |
| `boxes` | `s.arrow`, `s.label`, `sco.class`, `sco.label`, `sco.match` | | `plabels.boxes.draw` | |
| `cellipse` | `s.class`, `s.distri`, `triangle.class` | `ellipseSize` | | |
| `cgrid` | `s.arrow`, `s.class`, `s.chull`, `s.corcircle`, `s.distri`, `s.image`, `s.kde2d`, `s.label`, `s.logo`, `s.match`, `s.traject`, `s.value`, `sco.boxplot`, `sco.class`, `sco.distri`, `sco.gauss`, `sco.label`, `sco.match` | | `pgrid.nint` | both play on the grid mesh, but they are not strictly equivalent |
| `clabel` | `s.arrow`, `s.class`, `s.chull`, `s.corcircle`, `s.distri`, `s.kde2d`, `s.label`, `s.match`, `s.traject`, `sco.boxplot`, `sco.class`, `sco.distri`, `sco.gauss`, `sco.label`, `sco.match`, `triangle.plot` | | `plabels.cex` | |
| `clabel` | `table.dist` | | | `axis.text = list()` `lattice` parameter |
| `clabel.col` | `table.cont`, `table.paint`, `table.value` | | | `axis.text = list()` `lattice` parameter |
| `clabel.row` | `table.cont`, `table.paint`, `table.value` | | | `axis.text = list()` `lattice` parameter |
| `clegend` | `s.value`, `table.cont`, `table.value` | | `plegend.size` `ppoints.cex` | parameters setting the legend size |
| `clegend` | `table.paint` | | `plegend.size` | |
| `clogo` | `s.logo` | | `ppoints.cex` | |
| `cneig` | `s.image`, `s.kde2d`, `s.label`, `s.logo`, `s.value` | | `pnb.edge.lwd` | |
| `col.labels` | `table.cont`, `table.paint`, `table.value` | `labelsy` | | |
| `contour` | `s.arrow`, `s.class`, `s.chull`, `s.distri`, `s.image`, `s.kde2d`, `s.label`, `s.logo`, `s.match`, `s.traject`, `s.value` | `Sp` | | a `Sp` object |
| `contour.plot` | `s.image` | `region` | | |
| `cpoints`, `cpoint` | `s.arrow`, `s.class`, `s.chull`, `s.distri`, `s.kde2d`, `s.label`, `s.match`, `s.traject`, `s.value`, `sco.class`, `sco.label`, `sco.match`, `triangle.class`, `triangle.plot` | | `ppoints.cex` | |
| `csize` | `s.value`, `table.cont`, `table.dist`, `table.paint`, `table.value` | `ppoints.cex` | | |
| `csize` | `sco.distri` | `sdSize` | | |
| `cstar` | `s.class`, `s.distri`, `triangle.class` | `starSize` | | |
| `csub` | `s.arrow`, `s.chull`, `s.class`, `s.corcircle`, `s.distri`, `s.image`, `s.kde2d`, `s.label`, `s.logo`, `s.match`, `s.traject`, `s.value`, `sco.boxplot`, `sco.class`, `sco.distri`, `sco.gauss`, `sco.label`, `sco.match`, `triangle.class`, `triangle.plot` | | `psub.cex` | |
| `draw.line` | `triangle.biplot`, `triangle.class`, `triangle.plot` | | `pgrid.draw` | |
| `edge` | `s.arrow`, `s.match`, `s.traject` | | `parrows.length` | setting the length of the arrows to 0 is equivalent to `edge = FALSE` |
| `grid` | `s.arrow`, `s.chull`, `s.class`, `s.corcircle`, `s.distri`, `s.image`, `s.kde2d`, `s.label`, `s.logo`, `s.match`, `s.traject`, `s.value`, `sco.boxplot`, `sco.class`, `sco.distri`, `sco.gauss`, `sco.label`, `sco.match`, `table.cont`, `table.dist`, `table.value` | | `pgrid.draw` | |
| `horizontal` | `sco.class`, `sco.gauss`, `sco.label`, `sco.match` | | `p1d.horizontal` | |
| `image.plot` | `s.image` | `contour` | | |
| `includeorigin`, `include.origin` | `s.arrow`, `s.chull`, `s.class`, `s.distri`, `s.image`, `s.kde2d`, `s.label`, `s.logo`, `s.match`, `s.traject`, `s.value`, `sco.boxplot`, `sco.class`, `sco.distri`, `sco.gauss`, `sco.label`, `sco.match` | | `porigin.include` | |
| `kgrid` | `s.image` | `gridsize` | | |
| `klogo` | `s.logo` | | | no correspondence |
| `labeltriangle` | `triangle.class` , `triangle.plot` | | | no correspondence |
| `legen` | `sco.gauss` | `labelplot` | | |
| `neig` | `s.image`, `s.kde2d`, `s.label`, `s.logo`, `s.value` | `nbobject` | | a `nb` object |
| `optchull` | `s.chull` | `chullSize` | | |
| `origin` | `s.arrow`, `s.chull`, `s.class`, `s.corcircle`, `s.distri`, `s.image`, `s.kde2d`, `s.label`, `s.logo`, `s.match`, `s.traject`, `s.value`, `sco.boxplot`, `sco.class`, `sco.distri`, `sco.gauss`, `sco.label`, `sco.match` | | `porigin.origin` | |
| `pch` | `s.arrow`, `s.chull`, `s.class`, `s.distri`, `s.kde2d`, `s.label`, `s.match`, `s.traject`, `s.value`, `sco.boxplot`, `sco.class`, `sco.label`, `sco.match`, `triangle.class`, `triangle.plot`, `table.cont` | | `ppoints.pch` | |
| `pixmap` | `s.arrow`, `s.chull`, `s.class`, `s.distri`, `s.image`, `s.kde2d`, `s.label`, `s.logo`, `s.match`, `s.traject`, `s.value` | | | no correspondence |
| `pos.lab` | `sco.class`, `sco.label`, `sco.match` | | `p1d.labpos` | |
| `possub` | `s.arrow`, `s.chull`, `s.class`, `s.corcircle`, `s.distri`, `s.image`, `s.kde2d`, `s.label`, `s.logo`, `s.match`, `s.traject`, `s.value`, `sco.class`, `sco.gauss`, `sco.label`, `sco.match`, `triangle.class`, `triangle.plot` | | `psub.pos` | |
| `rectlogo` | `s.logo` | `rect` | | |
| `reverse` | `sco.class`, `sco.gauss`, `sco.label`, `sco.match` | | `p1d.reverse` | |
| `row.labels` | `table.cont`, `table.paint`, `table.value` | `labelsx` | | |
| `scale` | `triangle.class`, `triangle.plot` | `adjust` | | |
| `show.position` | `triangle.class`, `triangle.plot` | `showposition` | | |
| `sub` | `s.arrow`, `s.chull`, `s.class`, `s.corcircle`, `s.distri`, `s.image`, `s.kde2d`, `s.label`, `s.logo`, `s.match`, `s.traject`, `s.value`, `sco.boxplot`, `sco.class`, `sco.distri`, `sco.gauss`, `sco.label`, `sco.match`, `triangle.class`, `triangle.plot` | | `psub.text` | |
| `y.rank` | `sco.distri` | `yrank` | | |
| `zmax` | `s.value` | | | set to default max(abs(z)) |
| | | | | |
References
=============================