R › R help

Boxplot with Log10 and base-exponent axis

Classic

List

Threaded

8 messages Options

Luigi

Boxplot with Log10 and base-exponent axis

Dear all,

I would like to (i) produce boxplot graphs with axis in logarithm in base 10
and (ii) showing the values on the axis in 10^exponent format rather than
10E+exponent.

To illustrate with an example, I have some widely spread data that I chart
plot using boxplot() [figure on the left]; the log="y" option of boxplot()
I obtained the natural logarithm conversion of the data and the unfriendly
notation baseE+exponent [figure on the centre]; if I log10 the data I obtain
the desired plot, but the axis are showing only the exponent. [figure on the
right].

Can anybody help?

Best regards

Luigi Marongiu, MSc

########### EXAMPLE ############

# generationg random numbers

x<-runif(100, min=0, max=100000)

# create plot

par(mfrow = c(1,3))

#plotting in the left side

boxplot(x, xlab="Linear values")

#plotting in the centre

boxplot(x, log = "y", xlab="y axis logged")

# creating log10 values and plotting on the right side

Log.base10.x<-log10(x)

boxplot(Log.base10.x, xlab="LOG10 of data")

[[alternative HTML version deleted]]

______________________________________________
[hidden email] mailing list
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.

William Dunlap

Re: Boxplot with Log10 and base-exponent axis

The key is to supply an expression, not text, to the labels argument to axis.
See help("plotmath") for details. Here is an example:
x <- list(One=10^(sin(1:10)+5), Two=10^(cos(1:30)*2))
boxplot(x, log="y", yaxt="n")
ylim <- par("usr")[3:4]
log10AtY <- seq(ceiling(ylim[1]), floor(ylim[2]))
axis(side=2, at=10^log10AtY, lab=as.expression(lapply(log10AtY, function(y)bquote(10^.(y)))))

Bill Dunlap
Spotfire, TIBCO Software
wdunlap tibco.com

> -----Original Message-----
> From: [hidden email] [mailto:[hidden email]] On Behalf
> Of Luigi
> Sent: Friday, June 22, 2012 7:54 AM
> To: [hidden email]
> Subject: [R] Boxplot with Log10 and base-exponent axis
>
> Dear all,
>
> I would like to (i) produce boxplot graphs with axis in logarithm in base 10
> and (ii) showing the values on the axis in 10^exponent format rather than
> 10E+exponent.
>
>
>
> To illustrate with an example, I have some widely spread data that I chart
> plot using boxplot() [figure on the left]; the log="y" option of boxplot()
> I obtained the natural logarithm conversion of the data and the unfriendly
> notation baseE+exponent [figure on the centre]; if I log10 the data I obtain
> the desired plot, but the axis are showing only the exponent. [figure on the
> right].
>
>
>
> Can anybody help?
>
>
>
> Best regards
>
> Luigi Marongiu, MSc
>
>
>
> ########### EXAMPLE ############
>
> # generationg random numbers
>
> x<-runif(100, min=0, max=100000)
>
>
>
> # create plot
>
> par(mfrow = c(1,3))
>
>
>
> #plotting in the left side
>
> boxplot(x, xlab="Linear values")
>
>
>
> #plotting in the centre
>
> boxplot(x, log = "y", xlab="y axis logged")
>
>
>
> # creating log10 values and plotting on the right side
>
> Log.base10.x<-log10(x)
>
> boxplot(Log.base10.x, xlab="LOG10 of data")
>
>
>
>
>
>
> [[alternative HTML version deleted]]
>
> ______________________________________________
> [hidden email] mailing list
> 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.

______________________________________________
[hidden email] mailing list
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.

William Dunlap

Re: Boxplot with Log10 and base-exponent axis

You might prefer placing the tick marks with
log10AtY <- log10(axTicks(side=2))
log10AtY <- unique(round(log10AtY))
instead of the
ylim <- par("usr")[3:4]
log10AtY <- seq(ceiling(ylim[1]), floor(ylim[2]))
in my original example.

Bill Dunlap
Spotfire, TIBCO Software
wdunlap tibco.com

> -----Original Message-----
> From: [hidden email] [mailto:[hidden email]] On Behalf
> Of William Dunlap
> Sent: Friday, June 22, 2012 8:14 AM
> To: Luigi; [hidden email]
> Subject: Re: [R] Boxplot with Log10 and base-exponent axis
>
> The key is to supply an expression, not text, to the labels argument to axis.
> See help("plotmath") for details. Here is an example:
> x <- list(One=10^(sin(1:10)+5), Two=10^(cos(1:30)*2))
> boxplot(x, log="y", yaxt="n")
> ylim <- par("usr")[3:4]
> log10AtY <- seq(ceiling(ylim[1]), floor(ylim[2]))
> axis(side=2, at=10^log10AtY, lab=as.expression(lapply(log10AtY,
> function(y)bquote(10^.(y)))))
>
> Bill Dunlap
> Spotfire, TIBCO Software
> wdunlap tibco.com
>
> > -----Original Message-----
> > From: [hidden email] [mailto:[hidden email]] On Behalf
> > Of Luigi
> > Sent: Friday, June 22, 2012 7:54 AM
> > To: [hidden email]
> > Subject: [R] Boxplot with Log10 and base-exponent axis
> >
> > Dear all,
> >
> > I would like to (i) produce boxplot graphs with axis in logarithm in base 10
> > and (ii) showing the values on the axis in 10^exponent format rather than
> > 10E+exponent.
> >
> >
> >
> > To illustrate with an example, I have some widely spread data that I chart
> > plot using boxplot() [figure on the left]; the log="y" option of boxplot()
> > I obtained the natural logarithm conversion of the data and the unfriendly
> > notation baseE+exponent [figure on the centre]; if I log10 the data I obtain
> > the desired plot, but the axis are showing only the exponent. [figure on the
> > right].
> >
> >
> >
> > Can anybody help?
> >
> >
> >
> > Best regards
> >
> > Luigi Marongiu, MSc
> >
> >
> >
> > ########### EXAMPLE ############
> >
> > # generationg random numbers
> >
> > x<-runif(100, min=0, max=100000)
> >
> >
> >
> > # create plot
> >
> > par(mfrow = c(1,3))
> >
> >
> >
> > #plotting in the left side
> >
> > boxplot(x, xlab="Linear values")
> >
> >
> >
> > #plotting in the centre
> >
> > boxplot(x, log = "y", xlab="y axis logged")
> >
> >
> >
> > # creating log10 values and plotting on the right side
> >
> > Log.base10.x<-log10(x)
> >
> > boxplot(Log.base10.x, xlab="LOG10 of data")
> >
> >
> >
> >
> >
> >
> > [[alternative HTML version deleted]]
> >
> > ______________________________________________
> > [hidden email] mailing list
> > 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.
>
> ______________________________________________
> [hidden email] mailing list
> 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.

Martin Maechler

Re: Boxplot with Log10 and base-exponent axis

In reply to this post by William Dunlap

> The key is to supply an expression, not text, to the labels argument to axis.
> See help("plotmath") for details. Here is an example:

> x <- list(One=10^(sin(1:10)+5), Two=10^(cos(1:30)*2))
> boxplot(x, log="y", yaxt="n")
> ylim <- par("usr")[3:4]
> log10AtY <- seq(ceiling(ylim[1]), floor(ylim[2]))
> axis(side=2, at=10^log10AtY, lab=as.expression(lapply(log10AtY, function(y)bquote(10^.(y)))))

Yes, that's nice, and exactly the basic idea.

For a few years now, the eaxis() function in package "sfsmisc"
does (something like) this (and a bit more) even more nicely :

install.packages("sfsmisc")
require("sfsmisc")

x <- list(One=10^(sin(1:10)+5), Two=10^(cos(1:30)*2))
boxplot(x, log="y", yaxt="n")
eaxis(2)

--
Martin Maechler, ETH Zurich

> Bill Dunlap
> Spotfire, TIBCO Software
> wdunlap tibco.com

> > -----Original Message-----
> > From: [hidden email] [mailto:[hidden email]] On Behalf
> > Of Luigi
> > Sent: Friday, June 22, 2012 7:54 AM
> > To: [hidden email]
> > Subject: [R] Boxplot with Log10 and base-exponent axis
> >
> > Dear all,
> >
> > I would like to (i) produce boxplot graphs with axis in logarithm in base 10
> > and (ii) showing the values on the axis in 10^exponent format rather than
> > 10E+exponent.
> >
> >
> >
> > To illustrate with an example, I have some widely spread data that I chart
> > plot using boxplot() [figure on the left]; the log="y" option of boxplot()
> > I obtained the natural logarithm conversion of the data and the unfriendly
> > notation baseE+exponent [figure on the centre]; if I log10 the data I obtain
> > the desired plot, but the axis are showing only the exponent. [figure on the
> > right].
> >
> >
> >
> > Can anybody help?
> >
> >
> >
> > Best regards
> >
> > Luigi Marongiu, MSc
> >
> >
> >
> > ########### EXAMPLE ############
> >
> > # generationg random numbers
> >
> > x<-runif(100, min=0, max=100000)
> >
> >
> >
> > # create plot
> >
> > par(mfrow = c(1,3))
> >
> >
> >
> > #plotting in the left side
> >
> > boxplot(x, xlab="Linear values")
> >
> >
> >
> > #plotting in the centre
> >
> > boxplot(x, log = "y", xlab="y axis logged")
> >
> >
> >
> > # creating log10 values and plotting on the right side
> >
> > Log.base10.x<-log10(x)
> >
> > boxplot(Log.base10.x, xlab="LOG10 of data")
> >
> >
> >
> >
> >
> >
> > [[alternative HTML version deleted]]
> >
> > ______________________________________________
> > [hidden email] mailing list
> > 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.

Luigi

Re: Boxplot with Log10 and base-exponent axis

In reply to this post by William Dunlap

Thank you!
This works good. I understand that the value are now in Log10 scale,
although I did not understand what is happening at line 4 of your script.
I would like to ask how can I change the y limits since they now depend on
par("user"). What if I'd like y limits extending from 1000 to 1 000 000?
I've tried to modify ylim (line 3) and ceiling (line 4) but I obtained
errors.
Best wishes,
Luigi

############### UPDATED EXAMPLE ############################################
# generationg random numbers
x<-runif(100, min=0, max=100000)

#plotting
boxplot(x, log = "y", yaxt="n")
ylim <- par("usr")[3:4]

log10AtY <- seq(ceiling(ylim[1]), floor(ylim[2]))

axis(side=2, at=10^log10AtY, lab=as.expression(lapply(log10AtY,
function(y)bquote(10^.(y)))))

############################################################################
#

------------------ Your Response
--------------------------------------------
The key is to supply an expression, not text, to the labels argument to
axis.
See help("plotmath") for details. Here is an example:
1 x <- list(One=10^(sin(1:10)+5), Two=10^(cos(1:30)*2))
2 boxplot(x, log="y", yaxt="n")
3 ylim <- par("usr")[3:4]
4 log10AtY <- seq(ceiling(ylim[1]), floor(ylim[2]))
5 axis(side=2, at=10^log10AtY, lab=as.expression(lapply(log10AtY,
function(y)bquote(10^.(y)))))

Bill Dunlap
Spotfire, TIBCO Software
wdunlap tibco.com

> -----Original Message-----
> From: [hidden email] [mailto:[hidden email]]
On Behalf
> Of Luigi
> Sent: Friday, June 22, 2012 7:54 AM
> To: [hidden email]
> Subject: [R] Boxplot with Log10 and base-exponent axis
>
> Dear all,
>
> I would like to (i) produce boxplot graphs with axis in logarithm in base
10
> and (ii) showing the values on the axis in 10^exponent format rather than
> 10E+exponent.
>
>
>
> To illustrate with an example, I have some widely spread data that I chart
> plot using boxplot() [figure on the left]; the log="y" option of
boxplot()
> I obtained the natural logarithm conversion of the data and the unfriendly
> notation baseE+exponent [figure on the centre]; if I log10 the data I
obtain
> the desired plot, but the axis are showing only the exponent. [figure on
the

> right].
>
>
>
> Can anybody help?
>
>
>
> Best regards
>
> Luigi Marongiu, MSc
>
>
>
> ########### EXAMPLE ############
>
> # generationg random numbers
>
> x<-runif(100, min=0, max=100000)
>
>
>
> # create plot
>
> par(mfrow = c(1,3))
>
>
>
> #plotting in the left side
>
> boxplot(x, xlab="Linear values")
>
>
>
> #plotting in the centre
>
> boxplot(x, log = "y", xlab="y axis logged")
>
>
>
> # creating log10 values and plotting on the right side
>
> Log.base10.x<-log10(x)
>
> boxplot(Log.base10.x, xlab="LOG10 of data")
>
>
>
>
>
>
> [[alternative HTML version deleted]]
>
> ______________________________________________
> [hidden email] mailing list
> 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.

______________________________________________
[hidden email] mailing list
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.

William Dunlap

Re: Boxplot with Log10 and base-exponent axis

To change the y-limits you need to add ylim=c(min,max)
to the call to boxplot. The limits there should be
in the units of the y variable (not its log10). par("usr")[3:4]
report the y-limits on the log10 scale (they will be expanded
a bit from your desired limits so the plot does not extend
all the way to edge).

> boxplot(abs(tan(1:100)), log="y", ylim=c(.001, 1000))
> par("usr")[3:4]
[1] -3.24 3.24
> 10 ^ par("usr")[3:4]
[1] 5.754399e-04 1.737801e+03

Bill Dunlap
Spotfire, TIBCO Software
wdunlap tibco.com

> -----Original Message-----
> From: Luigi [mailto:[hidden email]]
> Sent: Saturday, June 23, 2012 8:49 AM
> To: William Dunlap
> Cc: 'Martin Maechler'; [hidden email]
> Subject: RE: [R] Boxplot with Log10 and base-exponent axis
>
> Thank you!
> This works good. I understand that the value are now in Log10 scale,
> although I did not understand what is happening at line 4 of your script.
> I would like to ask how can I change the y limits since they now depend on
> par("user"). What if I'd like y limits extending from 1000 to 1 000 000?
> I've tried to modify ylim (line 3) and ceiling (line 4) but I obtained
> errors.
> Best wishes,
> Luigi
>
>
> ############### UPDATED EXAMPLE ############################################
> # generationg random numbers
> x<-runif(100, min=0, max=100000)
>
>
> #plotting
> boxplot(x, log = "y", yaxt="n")
> ylim <- par("usr")[3:4]
>
> log10AtY <- seq(ceiling(ylim[1]), floor(ylim[2]))
>
> axis(side=2, at=10^log10AtY, lab=as.expression(lapply(log10AtY,
> function(y)bquote(10^.(y)))))
>
> ############################################################################
> #
>
>
> ------------------ Your Response
> --------------------------------------------
> The key is to supply an expression, not text, to the labels argument to
> axis.
> See help("plotmath") for details. Here is an example:
> 1 x <- list(One=10^(sin(1:10)+5), Two=10^(cos(1:30)*2))
> 2 boxplot(x, log="y", yaxt="n")
> 3 ylim <- par("usr")[3:4]
> 4 log10AtY <- seq(ceiling(ylim[1]), floor(ylim[2]))
> 5 axis(side=2, at=10^log10AtY, lab=as.expression(lapply(log10AtY,
> function(y)bquote(10^.(y)))))
>
> Bill Dunlap
> Spotfire, TIBCO Software
> wdunlap tibco.com
>
> > -----Original Message-----
> > From: [hidden email] [mailto:[hidden email]]
> On Behalf
> > Of Luigi
> > Sent: Friday, June 22, 2012 7:54 AM
> > To: [hidden email]
> > Subject: [R] Boxplot with Log10 and base-exponent axis
> >
> > Dear all,
> >
> > I would like to (i) produce boxplot graphs with axis in logarithm in base
> 10
> > and (ii) showing the values on the axis in 10^exponent format rather than
> > 10E+exponent.
> >
> >
> >
> > To illustrate with an example, I have some widely spread data that I chart
> > plot using boxplot() [figure on the left]; the log="y" option of
> boxplot()
> > I obtained the natural logarithm conversion of the data and the unfriendly
> > notation baseE+exponent [figure on the centre]; if I log10 the data I
> obtain
> > the desired plot, but the axis are showing only the exponent. [figure on
> the
> > right].
> >
> >
> >
> > Can anybody help?
> >
> >
> >
> > Best regards
> >
> > Luigi Marongiu, MSc
> >
> >
> >
> > ########### EXAMPLE ############
> >
> > # generationg random numbers
> >
> > x<-runif(100, min=0, max=100000)
> >
> >
> >
> > # create plot
> >
> > par(mfrow = c(1,3))
> >
> >
> >
> > #plotting in the left side
> >
> > boxplot(x, xlab="Linear values")
> >
> >
> >
> > #plotting in the centre
> >
> > boxplot(x, log = "y", xlab="y axis logged")
> >
> >
> >
> > # creating log10 values and plotting on the right side
> >
> > Log.base10.x<-log10(x)
> >
> > boxplot(Log.base10.x, xlab="LOG10 of data")
> >
> >
> >
> >
> >
> >
> > [[alternative HTML version deleted]]
> >
> > ______________________________________________
> > [hidden email] mailing list
> > 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.

Steve Taylor

Re: Boxplot with Log10 and base-exponent axis

In reply to this post by Martin Maechler

This (and William's solution) is so good, I must ask:
Is there a good reason why this is not the default functionality in the graphics package?

The default displays the number 1 as 1+e00 which is hideous!

-----Original Message-----
From: [hidden email] [mailto:[hidden email]] On Behalf Of Martin Maechler
Sent: Saturday, 23 June 2012 4:57a
To: Luigi; William Dunlap
Cc: [hidden email]
Subject: Re: [R] Boxplot with Log10 and base-exponent axis

> The key is to supply an expression, not text, to the labels argument to axis.
> See help("plotmath") for details. Here is an example:

> x <- list(One=10^(sin(1:10)+5), Two=10^(cos(1:30)*2))
> boxplot(x, log="y", yaxt="n")
> ylim <- par("usr")[3:4]
> log10AtY <- seq(ceiling(ylim[1]), floor(ylim[2]))
> axis(side=2, at=10^log10AtY, lab=as.expression(lapply(log10AtY, function(y)bquote(10^.(y)))))

Yes, that's nice, and exactly the basic idea.

For a few years now, the eaxis() function in package "sfsmisc"
does (something like) this (and a bit more) even more nicely :

install.packages("sfsmisc")
require("sfsmisc")

x <- list(One=10^(sin(1:10)+5), Two=10^(cos(1:30)*2))
boxplot(x, log="y", yaxt="n")
eaxis(2)

--
Martin Maechler, ETH Zurich

______________________________________________
[hidden email] mailing list
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.

Luigi

Re: Boxplot with Log10 and base-exponent axis

In reply to this post by William Dunlap

Dear Mr Dunlap,
Your solution works really fine. Thank you for your time,
Best wishes,
Luigi

-----Original Message-----
From: William Dunlap [mailto:[hidden email]]
Sent: 23 June 2012 18:48
To: Luigi
Cc: 'Martin Maechler'; [hidden email]
Subject: RE: [R] Boxplot with Log10 and base-exponent axis

To change the y-limits you need to add ylim=c(min,max)
to the call to boxplot. The limits there should be
in the units of the y variable (not its log10). par("usr")[3:4]
report the y-limits on the log10 scale (they will be expanded
a bit from your desired limits so the plot does not extend
all the way to edge).

> boxplot(abs(tan(1:100)), log="y", ylim=c(.001, 1000))
> par("usr")[3:4]
[1] -3.24 3.24
> 10 ^ par("usr")[3:4]
[1] 5.754399e-04 1.737801e+03

Bill Dunlap
Spotfire, TIBCO Software
wdunlap tibco.com

############################################

> # generationg random numbers
> x<-runif(100, min=0, max=100000)
>
>
> #plotting
> boxplot(x, log = "y", yaxt="n")
> ylim <- par("usr")[3:4]
>
> log10AtY <- seq(ceiling(ylim[1]), floor(ylim[2]))
>
> axis(side=2, at=10^log10AtY, lab=as.expression(lapply(log10AtY,
> function(y)bquote(10^.(y)))))
>
>

############################################################################

> #
>
>
> ------------------ Your Response
> --------------------------------------------
> The key is to supply an expression, not text, to the labels argument to
> axis.
> See help("plotmath") for details. Here is an example:
> 1 x <- list(One=10^(sin(1:10)+5), Two=10^(cos(1:30)*2))
> 2 boxplot(x, log="y", yaxt="n")
> 3 ylim <- par("usr")[3:4]
> 4 log10AtY <- seq(ceiling(ylim[1]), floor(ylim[2]))
> 5 axis(side=2, at=10^log10AtY, lab=as.expression(lapply(log10AtY,
> function(y)bquote(10^.(y)))))
>
> Bill Dunlap
> Spotfire, TIBCO Software
> wdunlap tibco.com
>
> > -----Original Message-----
> > From: [hidden email] [mailto:[hidden email]]
> On Behalf
> > Of Luigi
> > Sent: Friday, June 22, 2012 7:54 AM
> > To: [hidden email]
> > Subject: [R] Boxplot with Log10 and base-exponent axis
> >
> > Dear all,
> >
> > I would like to (i) produce boxplot graphs with axis in logarithm in

base
> 10
> > and (ii) showing the values on the axis in 10^exponent format rather
than
> > 10E+exponent.
> >
> >
> >
> > To illustrate with an example, I have some widely spread data that I
chart
> > plot using boxplot() [figure on the left]; the log="y" option of
> boxplot()
> > I obtained the natural logarithm conversion of the data and the
unfriendly

> > notation baseE+exponent [figure on the centre]; if I log10 the data I
> obtain
> > the desired plot, but the axis are showing only the exponent. [figure on
> the
> > right].
> >
> >
> >
> > Can anybody help?
> >
> >
> >
> > Best regards
> >
> > Luigi Marongiu, MSc
> >
> >
> >
> > ########### EXAMPLE ############
> >
> > # generationg random numbers
> >
> > x<-runif(100, min=0, max=100000)
> >
> >
> >
> > # create plot
> >
> > par(mfrow = c(1,3))
> >
> >
> >
> > #plotting in the left side
> >
> > boxplot(x, xlab="Linear values")
> >
> >
> >
> > #plotting in the centre
> >
> > boxplot(x, log = "y", xlab="y axis logged")
> >
> >
> >
> > # creating log10 values and plotting on the right side
> >
> > Log.base10.x<-log10(x)
> >
> > boxplot(Log.base10.x, xlab="LOG10 of data")
> >
> >
> >
> >
> >
> >
> > [[alternative HTML version deleted]]
> >
> > ______________________________________________
> > [hidden email] mailing list
> > 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.