R help - Nov 2007 - Writing a helper function that takes in the dataframe and variable names and then does a subset and plot

Hi,

I have a large dataframe than I'm writing functions to explore, and to 
reduce cut and paste I'm trying to write a function that does a subset 
and then a plot.

Firstly, I can write a wrapper around a plot:

plotwithfits <- function(formula, data, xylabels=c('','')) {
    xyplot(formula, data,  panel           function(x,y, ...) {
             panel.xyplot(x,y, ...)
             panel.abline(lm(y~x),lwd=2, ...)
             panel.loess(x,y,lwd=2,col.line='red', ...)
          },
          xlab = xylabels[1], ylab= xylabels[2])
}

plotwithfits(Latency ~ Stimulus.number | Subject.ID,eye_subsetted_data) 
# Works



However, I can't get it working if I try the same for a subset and plot:

explorebysubject <- 
function(xvar,yvar,data,condition=TRUE,xlim=c(-Inf,Inf),ylim=c(-Inf,Inf)) {
   
    temp_subset <- subset(data,
                          
subset=condition&xvar>xlim[1]&xvar<xlim[2]&yvar>ylim[1]&yvar<ylim[2],
                          select=c(Group,Subject.ID,xvar,yvar)
                          )              
                         
    plotwithfits(xvar~yvar | Subject.ID,temp_subset)
}

explorebysubject(Latency,Primary.gain,eye,Analysis.type == 'reflexive') 
# Doesn't work as can't find 'Analysis.type', 'Latency',
etc

I can see why it doesn't work, however, I've looked at substitute, 
deparse, etc, without much luck.

Is there a something simple I could do to fix this up? Is there any 
material I should be reading?

Using the arguments eye$Latency, eye$Primary.gain, etc would partially 
solve this problem, however I would prefer to avoid this if possible.

One unclean way that does work is constructing strings and then 
evaluating them. However, as I am planning to write several similar 
functions I would prefer to avoid this:

explorebysubject <- 
function(xvar,yvar,dataset,condition,xlim=c(-Inf,Inf),ylim=c(-Inf,Inf)) {
    # xvar - variable to plot along x-axis
    # yvar - variable to plot along y-axis
    # the dataset to use
    # condition - used to provide extra conditions on the data selected
    # xlim, ylim - optional limits of data to select. i.e. xlim=c(0,1)

    # Generate command to select appropriate data
    cmd <- 
paste("subset(",dataset,",",condition,"&",xvar,">",xlim[1],"&",xvar,"<",xlim[2],"&",yvar,">",ylim[1],"&",yvar,"<",ylim[2],",select=c(",xvar,",",yvar,",Group,Subject.ID))")

    temp_subset <- eval(parse(text=cmd))
   
    #generate plot command                     
    cmd <- 
paste("plotwithfits(",xvar,"~",yvar,"|Subject.ID,temp_subset)")
    eval(parse(text=cmd))
   
    }

explorebysubject('Latency','Primary.gain','eye',"Analysis.type
==
'reflexive'",xlim=c(90,500),ylim=c(0.5,1.5)) # Works

Thanks.

Cheers,
Daniel

Dear Daniel,

May I point you to Thomas Lumley's paper in R News 2001-3
("Programmer?s
Niche: Macros in R\n Overcoming R?s virtues) and to the defmacro utility
of the gtools package ?

HTH

					Emmanuel Charpentier


Daniel Myall a ?crit :
> Hi,
> 
> I have a large dataframe than I'm writing functions to explore, and to 
> reduce cut and paste I'm trying to write a function that does a subset 
> and then a plot.
> 
> Firstly, I can write a wrapper around a plot:
> 
> plotwithfits <- function(formula, data,
xylabels=c('','')) {
>     xyplot(formula, data,  panel >           function(x,y, ...) {
>              panel.xyplot(x,y, ...)
>              panel.abline(lm(y~x),lwd=2, ...)
>              panel.loess(x,y,lwd=2,col.line='red', ...)
>           },
>           xlab = xylabels[1], ylab= xylabels[2])
> }
> 
> plotwithfits(Latency ~ Stimulus.number | Subject.ID,eye_subsetted_data) 
> # Works
> 
> 
> 
> However, I can't get it working if I try the same for a subset and
plot:
> 
> explorebysubject <- 
> function(xvar,yvar,data,condition=TRUE,xlim=c(-Inf,Inf),ylim=c(-Inf,Inf)) {
>    
>     temp_subset <- subset(data,
>                           
>
subset=condition&xvar>xlim[1]&xvar<xlim[2]&yvar>ylim[1]&yvar<ylim[2],
>                           select=c(Group,Subject.ID,xvar,yvar)
>                           )              
>                          
>     plotwithfits(xvar~yvar | Subject.ID,temp_subset)
> }
> 
> explorebysubject(Latency,Primary.gain,eye,Analysis.type ==
'reflexive')
> # Doesn't work as can't find 'Analysis.type',
'Latency', etc
> 
> I can see why it doesn't work, however, I've looked at substitute, 
> deparse, etc, without much luck.
> 
> Is there a something simple I could do to fix this up? Is there any 
> material I should be reading?
> 
> Using the arguments eye$Latency, eye$Primary.gain, etc would partially 
> solve this problem, however I would prefer to avoid this if possible.
> 
> One unclean way that does work is constructing strings and then 
> evaluating them. However, as I am planning to write several similar 
> functions I would prefer to avoid this:
> 
> explorebysubject <- 
> function(xvar,yvar,dataset,condition,xlim=c(-Inf,Inf),ylim=c(-Inf,Inf)) {
>     # xvar - variable to plot along x-axis
>     # yvar - variable to plot along y-axis
>     # the dataset to use
>     # condition - used to provide extra conditions on the data selected
>     # xlim, ylim - optional limits of data to select. i.e. xlim=c(0,1)
> 
>     # Generate command to select appropriate data
>     cmd <- 
>
paste("subset(",dataset,",",condition,"&",xvar,">",xlim[1],"&",xvar,"<",xlim[2],"&",yvar,">",ylim[1],"&",yvar,"<",ylim[2],",select=c(",xvar,",",yvar,",Group,Subject.ID))")
> 
>     temp_subset <- eval(parse(text=cmd))
>    
>     #generate plot command                     
>     cmd <- 
>
paste("plotwithfits(",xvar,"~",yvar,"|Subject.ID,temp_subset)")
>     eval(parse(text=cmd))
>    
>     }
> 
>
explorebysubject('Latency','Primary.gain','eye',"Analysis.type
==
> 'reflexive'",xlim=c(90,500),ylim=c(0.5,1.5)) # Works
> 
> Thanks.
> 
> Cheers,
> Daniel
> 
> ______________________________________________
> R-help at r-project.org 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.
>