[R] second legend in biplot

Thu Dec 10 10:36:26 CET 2020

Hallo Jim

It is OK, your discussion was helpful as I found another way how to present 
PCA results.

It is my job to tweek the code to suite my purpose.

Cheers
Petr

> -----Original Message-----
> From: Jim Lemon <drjimlemon using gmail.com>
> Sent: Thursday, December 10, 2020 9:57 AM
> To: PIKAL Petr <petr.pikal using precheza.cz>
> Subject: Re: [R] second legend in biplot
>
> Ah, gotcha. I thought you just wanted the fancy legend. Sorry, I haven't got
> time to rewrite functions at that moment.
>
> Jim
>
> On Thu, Dec 10, 2020 at 7:47 PM PIKAL Petr <petr.pikal using precheza.cz>
> wrote:
> >
> > Thank you Jim
> >
> > I am almost sure that biplot and prcomp itself cannot do point (text)
> > coding according to some factor.
> >
> > The closest way to what I want to achieve is to use result from prcomp
> > and add points into existing biplot, something like that
> >
> > PCA <- prcomp(iris[,-5], scale=T, center=T)
> > biplot(PCA)
> > points(PCA$x[,1:2], pch=20, col=iris$Species)
> > >
> > But maybe tweeking the code from fviz_pca_biplot could be easier as it
> > is possible to have separate legend for point colour and shape but
> > shape and colour is the same.
> > https://stackoverflow.com/questions/47482879/how-to-make-a-pretty-
> bipl
> > ot-in-r-without-using-external-packages
> >
> > I want shape and colour coded by different factors.
> >
> > Anyway, thanks for your effort.
> >
> > Cheers
> > Petr
> >
> > > -----Original Message-----
> > > From: Jim Lemon <drjimlemon using gmail.com>
> > > Sent: Wednesday, December 9, 2020 11:37 PM
> > > To: PIKAL Petr <petr.pikal using precheza.cz>
> > > Cc: R mailing list <r-help using r-project.org>
> > > Subject: Re: [R] second legend in biplot
> > >
> > > Hi Petr,
> > > Perhaps legendg in plotrix can help:
> > >
> > > legendg(-180,600,c("State","Crime"),lty=1,col=list("black","red"),pc
> > > h=3:4)
> > >
> > > Jim
> > >
> > > On Wed, Dec 9, 2020 at 10:31 PM PIKAL Petr <petr.pikal using precheza.cz>
> > > wrote:
> > > >
> > > > Thank you Jim.
> > > >
> > > > biplot can have distinct colours of points and arrows but not
> > > > points coded according to some factor. In simple example I
> > > > provided, points are coded according by 2 factors. What I would
> > > > like to achieve is to colour points and have their shapes coded by
> > > > second factor, which fviz_pca_biplot does, however it does not
> > > > present this second factor in the
> > > legend.
> > > >
> > > > Probably the last resort is to rewrite original code which I would
> > > > like to avoid as I am not so experienced in grid graphics.
> > > >
> > > > Best regards
> > > > Petr
> > > >
> > > > > -----Original Message-----
> > > > > From: Jim Lemon <drjimlemon using gmail.com>
> > > > > Sent: Monday, December 7, 2020 9:53 PM
> > > > > To: PIKAL Petr <petr.pikal using precheza.cz>
> > > > > Cc: R mailing list <r-help using r-project.org>
> > > > > Subject: Re: [R] second legend in biplot
> > > > >
> > > > > Hi Petr,
> > > > > Here's an attempt, using the example in biplot.princomp:
> > > > >
> > > > > biplot(princomp(USArrests))
> > > > > > par("usr")
> > > > > [1] -497.2263  624.8856 -497.2263  624.8856
> > > > > legend(-180,600,c("State","Crime"),lty=1,col=c("black","red"))
> > > > >
> > > > > Jim
> > > > >
> > > > > On Mon, Dec 7, 2020 at 6:23 PM PIKAL Petr
> > > > > <petr.pikal using precheza.cz>
> > > wrote:
> > > > > >
> > > > > > Dear all
> > > > > >
> > > > > > I try to make fviz_pca_biplot with 2 (or more) legends. Below
> > > > > > is data and the code, which gives one legend (colour) for
> > > > > > coating variable and correctly shows triangles and circles for
> > > > > > size variable. But this is not shown in the legend. Hopefully
> > > > > > somebody could
> > > help.
> > > > > >
> > > > > > And before you ask, I tried to contact maintainer about a week
> > > > > > ago but did not get response, therefore I try to post again here.
> > > > > >
> > > > > > Even an answer that current version of fviz_pca_biplot does
> > > > > > not support 2 legends with different point shapes is OK as it
> > > > > > will send me either to adapt source code or to stop trying
> impossible.
> > > > > >
> > > > > > Best regards
> > > > > > Petr
> > > > > >
> > > > > > library(factoextra)
> > > > > > library(FactoMineR)
> > > > > >
> > > > > > fit <- PCA(temp, quali.sup=c(9,10)) fviz_pca_biplot(fit,
> > > > > > col.ind = temp$coating, repel=T, col.var = "black", palette =
> > > > > > "lancet", invisible="quali", pointsize=5,
> > > > > > pointshape=temp$size, legend.title = list(col = "Coating",
> > > > > > shape="Size"), xlim=c(-6,6), title="Instillation results")
> > > > > >
> > > > > > temp <- structure(list(leukocyte28 = c(96875L, 73438L, 68229L,
> > > > > > 94479L, 76563L, 141667L, 111042L, 93333L, 132083L, 103542L,
> > > > > > 61667L, 77708L ),
> > > > > > macrophage28 = c(60.29, 99.13, 97.04, 98.54, 98.46, 75.2,
> > > > > > 89.71, 98, 82, 98.83, 99.08, 98.54), pmn28 = c(38.58, 0.58,
> > > > > > 2.71, 0.92, 1, 24.25, 9.29, 1.5, 15.08, 0.92, 0.67, 1),
> > > > > > lymphocyte28 = c(1.13, 0.29, 0.25, 0.54, 0.54, 0.55, 1, 0.5, 2.92,
> 0.25, 0.25, 0.46),
> > > > > >     leukocyte3 = c(186042L, 111250L, 114375L, 111146L, 98854L,
> > > > > >     156250L, 250625L, 183125L, 202917L, 161875L, 184792L,
> 128333L
> > > > > >     ), macrophage3 = c(53.88, 95.96, 98.29, 98.92, 98.92, 78.3,
> > > > > >     82.33, 97.83, 84.79, 97.25, 97.75, 98.46), pmn3 = c(44.75,
> > > > > >     3.46, 1.29, 0.67, 0.71, 20.4, 16.67, 1.92, 14.04, 1.92, 1.67,
> > > > > >     1.21), lymphocyte3 = c(1.38, 0.58, 0.42, 0.42, 0.38, 1.3,
> > > > > >     1, 0.25, 1.17, 0.83, 0.58, 0.33), coating = structure(c(3L,
> > > > > >     3L, 3L, 1L, 7L, 1L, 2L, 5L, 4L, 6L, 3L, 3L), .Label = 
> > > > > > c("alumina",
> > > > > >     "both", "none", "phosphate", "silica", "tungsten", "zirconia"
> > > > > >     ), class = "factor"), size = structure(c(1L, 1L, 2L, 2L,
> > > > > >     2L, 1L, 2L, 1L, 2L, 1L, 2L, 2L), .Label = c("nano", 
> > > > > > "pigmentary"
> > > > > >     ), class = "factor")), class = "data.frame", row.names =
> > > > > > c(NA,
> > > > > > 12L))
> > > > > >
> > > > > > ______________________________________________
> > > > > > R-help using r-project.org mailing list -- To UNSUBSCRIBE and more,
> > > > > > see https://stat.ethz.ch/mailman/listinfo/r-help
> > > > > > PLEASE do read the posting guide
> > > > > > http://www.R-project.org/posting-guide.html
> > > > > > and provide commented, minimal, self-contained, reproducible
> code.