Howto Plot ROC curve in R with only known SN/PPV/Cutoff info

Question

Given such data:

#Cutpoint         SN   (1-PPV)
5       0.56    0.01
7       0.78    0.19
9       0.91    0.58

How can I plot ROC curve with R that produce similar result like the attached ?

I know ROCR package but it doesn't take such input.

Answer 1

If you just want to create the plot (without that silly interpolation spline between points) then just plot the data you give in the standard way, prepending a point at (0,0) and appending one at (1,1) to give the end points of the curve.

## your data with different labels
dat <- data.frame(cutpoint = c(5, 7, 9),
                  TPR = c(0.56, 0.78, 0.91),
                  FPR = c(0.01, 0.19, 0.58))

## plot version 1    
op <- par(xaxs = "i", yaxs = "i")
plot(TPR ~ FPR, data = dat, xlim = c(0,1), ylim = c(0,1), type = "n")
with(dat, lines(c(0, FPR, 1), c(0, TPR, 1), type = "o", pch = 25, bg = "black"))
text(TPR ~ FPR, data = dat, pos = 3, labels = dat$cutpoint)
abline(0, 1)
par(op)

To explain the code: The first plot() call sets up the plotting region, without doing an plotting at all. Note that I force the plot to cover the range (0,1) in both axes. The par() call tells R to plot axes that cover the range of the data - the default extends them by 4 percent of the range on each axis.

The next line, with(dat, lines(....)) draws the ROC curve and here we prepend and append the points at (0,0) and (1,1) to give the full curve. Here I use type = "o" to give both points and lines overplotted, the points are represented by character 25 which allows it to be filled with a colour, here black.

Then I add labels to the points using text(....); the pos argument is used to position the label away from the actual plotting coordinates. I take the labels from the cutpoint object in the data frame.

The abline() call draws the 1:1 line (here the 0, and 1 mean an intercept of 0 and a slope of 1 respectively.

The final line resets the plotting parameters to the defaults we saved in op prior to plotting (in the first line).

The resulting plot looks like this:

It isn't an exact facsimile and I prefer the plot using the default for the axis ranges(adding 4 percent):

plot(TPR ~ FPR, data = dat, xlim = c(0,1), ylim = c(0,1), type = "n")
with(dat, lines(c(0, FPR, 1), c(0, TPR, 1), type = "o", pch = 25, bg = "black"))
text(TPR ~ FPR, data = dat, pos = 3, labels = dat$cutpoint)
abline(0, 1)

Again, not a true facsimile but close.