How do I scales the axes to the larger vector when plotting two ecdfs for comparison in R?

Question

Initially I start out with 2 vectors (subsets of my data). I run ecdf on both, plot them in the same plot for ease of comparison. All of that is fine but what I need to know is how to make the function work universally for any pair of vectors, so I can just enter in the vector and the function will work. Like if I were to call the larger vector second I want the axes of the plot to automatically scale for the larger vector, regardless of order called so that no data is lost.

I have included a set up to use the iris data just in case.

data=iris

virg<-subset(iris, iris$Species=="virginica"); virg
virg_pl<-virg$Petal.Length; virg_pl

versi<-subset(iris, iris$Species=="versicolor"); versi
versi_pl<-versi$Petal.Length; versi_pl

here is what i have:

twoecdfsoner<-function(x,y,z){
       ecdf1<-ecdf(x)
       ecdf2<-ecdf(y)
    plot(ecdf1,xlab=head(z,n=1),
        ylab="cumulative relative frequency",
        lty=1,pch=".",
        main="",
        do.point=FALSE,
        verticals=TRUE,xlim=c(min (y),max (x)))

    plot(ecdf2,verticals=TRUE, 
        do.points=FALSE,
        lty=3,pch=".", 
        add=TRUE, xlim=c(min (y),max (x)))

    legend("right","center",
    legend=c(deparse(substitute(x)),
        deparse(substitute(y))),
        lty=c(1,3),cex=0.8)
    }
    twoecdfsoner(virg_pl,versi_pl,"inches")

It seems like i could write a conditional statement but i get this error:

Warning messages:
1: In x > y :
  longer object length is not a multiple of shorter object length
2: In x > y :
  longer object length is not a multiple of shorter object length
3: In x > y :
  longer object length is not a multiple of shorter object length

so far i have tried

xlim=c(min (y),max (x)))
xlim=range(c(x),c(y)
xlim=pmax(x,y)

and writing conditional statements

I would also like the solid line to code for that larger vector. If anyone has any suggestions it would be greatly appreciated.

@42- after reading up a bit I thought I could do a conditional statement, this seems to also work. Do have any critique of running the code this way?

twoecdfsoner<-function(x,y,z){
ecdf_1 <- plot(ecdf(x),
    verticals=TRUE,
    pch=".",
    main="",
    do.points=FALSE,
    lty=ifelse(max(x)>max(y), c(1), c(3)),
    xlab=head(z,n=1),
    ylab="Cumulative relative frequency", 
    xlim=range(x,y))

ecdf_2 <- lines(ecdf(y), 
    verticals=TRUE, 
    do.points=FALSE,
    lty=ifelse(max(y)>max(x), c(1), c(3)), 
    pch=".")

legend_text<-
if (max(x)>max(y)){
    legend=c(deparse(substitute(x)), deparse(substitute(y)))
} else {max(y)>max(x)
    legend=c(deparse(substitute(y)), deparse(substitute(x)))
}


legend("right",
    legend=legend_text,
    lty=c(1,3))
}

twoecdfsoner(virg_pl,versi_pl,"inches")

@42 I see. No error, but the graph is scaled poorly. Got it. — G5W, Oct 07 '17 at 21:36
I think `xlim=c(min(c(x,y)), max(c(x,y))))` fixes your problem. — G5W, Oct 07 '17 at 21:39
Sadly, the scaling problem is still there according to which vector is called first. Maybe the iris data was a bad choice as an example, let me see if I can get my data up here. — m9000, Oct 07 '17 at 21:50

IRTFM · Accepted Answer · 2017-10-07T22:08:47.580

There's a problem with ecdf ... it hides the "x" arguments in an environment. Newbies then cannot find them.

> ecdf(versi$Petal.Length)
Empirical CDF 
Call: ecdf(versi$Petal.Length)
 x[1:19] =      3,    3.3,    3.5,  ...,      5,    5.1
> str(ecdf(versi$Petal.Length))
function (v)  
 - attr(*, "class")= chr [1:3] "ecdf" "stepfun" "function"
 - attr(*, "call")= language ecdf(versi$Petal.Length)

At this point it would be instructive to run all the examples in the help page:

?ecdf   # and probably also look at ?stepfun

Notice that the result is a function, so trying to derive a min and/or max with no argument will always fail. Once you do find how to access the environment of ecdf (see below) then you would probably want the min of the concatenated values of the two vectors, rather than assuming one has the min and the other has the max. Here's what's in the environments of the ecdf functions:

ls( environment(ecdf(versi_pl)) )
[1] "f"      "method" "nobs"   "x"      "y"      "yleft"  "yright"

 twoecdfsoner<-function(x,y,z){
       ecdf1<-ecdf(x);
       x1 <- environment(ecdf1)$x
       ecdf2<-ecdf(y); 
       x2 <- environment(ecdf2)$x
    plot(ecdf1,xlab=head(z,n=1),
        ylab="cumulative relative frequency",
        lty=1,pch=".",
        main="",
        do.point=FALSE,
        verticals=TRUE,xlim=c( min ( c(x1,x2) ),max ( c(x1,x2)) ) )

    plot(ecdf2,verticals=TRUE, 
        do.points=FALSE,
        lty=3,pch=".", 
        add=TRUE, xlim=c( min ( c(x1,x2) ), max ( c(x1,x2) )) ) 

    legend("right","center",
    legend=c(deparse(substitute(x)),
        deparse(substitute(y))),
        lty=c(1,3),cex=0.8)
    }
    twoecdfsoner(versi_pl, virg_pl,"inches")

How do I scales the axes to the larger vector when plotting two ecdfs for comparison in R?

1 Answers1