Skip to content

Plot question

4 messages · Alan Barnett, Marc Schwartz, Mark Difford +1 more

Alan Barnett
# 
I have some data consisting of multiple trials of an experiment with 
different values of an independent variable.  If I run
R> plot(var,result)
I get a scatterplot of result versus the independent variable var.
If I run
R> plot(as.factor(var),result)
I get a boxplot of the distribution of result for each value of var.  In 
this plot, each boxplot is labeled by the corresponding value of var, 
but the absissas are evenly spaced.
Is it possible to generate a boxplot with the absissas of each boxplot 
equal to the corresponding value of var?
Marc Schwartz
#
On Fri, 2007-12-14 at 08:16 -0500, Alan Barnett wrote:
If you want boxplots, you could do something like this:

CV <- rnorm(100)
FV <- sample(c(1, 3, 6), 100, replace = TRUE)

boxplot(CV ~ factor(FV, levels = seq(max(FV))))


This essentially 'fills in' the missing values (levels) of the factor
variable so that they are included in the plot.

HTH,

Marc Schwartz
Mark Difford
# 
Hi Alan,

Yes it is, but you need to do a bit work.  There are different approaches. 
Look at the at= option under

?bxp

and draw your boxplots with something like:

boxplot(y ~ as.numeric(as.factor(grp)), at=c(0.5, 2, 2.5, 3), xaxt="n", ...)
axis(side=1, at=c(0.5, 2, 2.5, 3), labels=c("0.5","2","2.5","3"))

This should do it, and is perhaps the easiest route.  There may be something
missing, because I don't have a concrete example.  Anyhow, you should be
able to fill in the gaps.

HTH,

Mark.
Alan Barnett wrote:

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Richard M. Heiberger
    

    

      
      #
    

  


  

      
The easiest way would be to use the HH package, which you can get from CRAN.

Marc's example is limited to integers on the X axis.  Using positioned(),
which is
an extension to ordered(), allows arbitrary values.  Building on that
example

require(HH)

CV <- rnorm(100)
FV <- sample(c(-1.4, 3.2, 5), 100, replace = TRUE)

bwplot(CV ~ positioned(FV))



-----Original Message-----
From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org] On
Behalf Of Alan Barnett
Sent: Friday, December 14, 2007 08:16 AM
To: R-Help
Subject: [R] Plot question

I have some data consisting of multiple trials of an experiment with 
different values of an independent variable.  If I run
R> plot(var,result)
I get a scatterplot of result versus the independent variable var.
If I run
R> plot(as.factor(var),result)
I get a boxplot of the distribution of result for each value of var.  In 
this plot, each boxplot is labeled by the corresponding value of var, 
but the absissas are evenly spaced.
Is it possible to generate a boxplot with the absissas of each boxplot 
equal to the corresponding value of var?