如何在泰勒图中标记点？

Question

The plotrix package has a function called taylor.diagram that plots two vectors - one representing data and the other model output.

Here is an example:

require(plotrix)
set.seed(10)
data  <- sort(runif(100, 8,12))
model <- sort(rnorm(100, 10, 4))
taylor.diagram(data, model)

And in this example I want to update the plot after improving the model:

model2 <- sort(rnorm(100, 10,2))
taylor.diagram(data, model2, add = TRUE)

To produce this:

How can I add labels such as "Model 1" and "Model 2" to identify these points? (Update: with the label position determined from the model values rather than done post-hoc)

Answer 1

A third solution is to create a modified version of the function taylor.diagram that includes a text label. In this case all there is to do is adding a parameter, say text , and after the call to points in the original function (2 lines before the closing braces), the line text(sd.f * R, sd.f * sin(acos(R)), labels=text, pos=3) .

taylor.diagram.modified <- function (ref, model, add = FALSE, col = "red", 
                                    pch = 19, pos.cor = TRUE, xlab = "", ylab = "", 
                                    main = "Taylor Diagram", show.gamma = TRUE, 
                                    ngamma = 3, gamma.col = 8, sd.arcs = 0, ref.sd = FALSE, 
                                    grad.corr.lines = c(0.2, 0.4, 0.6, 0.8, 0.9), pcex = 1, 
                                    cex.axis = 1, normalize = FALSE, mar = c(5, 4, 6, 6),
                                    text, ...) #the added parameter
{
    grad.corr.full <- c(0, 0.2, 0.4, 0.6, 0.8, 0.9, 0.95, 0.99,1)
    R <- cor(ref, model, use = "pairwise")
    sd.r <- sd(ref)
    sd.f <- sd(model)
    if (normalize) {

    ... #I didn't copy here the full function because it's quite long: to obtain it
    ... #simply call `taylor.diagram` in the console or `edit(taylor.diagram)`.

            }
            S <- (2 * (1 + R))/(sd.f + (1/sd.f))^2
        }
    }
    points(sd.f * R, sd.f * sin(acos(R)), pch = pch, col = col, 
           cex = pcex)
    text(sd.f * R, sd.f * sin(acos(R)),  #the line to add
         labels=text, cex = pcex, pos=3) #You can change the pos argument to your liking
    invisible(oldpar)
}

Then simply supply a label name in the text argument:

require(plotrix)
set.seed(10)
data  <- sort(runif(100, 8,12))
model <- sort(rnorm(100, 10, 4))
taylor.diagram.modified(data, model, text="Model 1")
model2 <- sort(rnorm(100, 10,2))
taylor.diagram.modified(data, model2, add = TRUE, text="Model 2")

Answer 2

Here are two approaches

1. example(taylor.diagram) shows a decent approach to placing a legend in the upper right corner (at 1.5*sd(data), 1.5*sd(data) ), but this would require different colors for the two points.

2. Another option would be to calculate the locations based on the equations from the original Taylor 2001 reference - or copy them from source code to the taylor.diagram function, near

    dy <- 1.1 # text offset coefficient sd.f <- sd(model) R <- cor(data, model, use = 'pairwise') x <- sd.f * R y <- sd.f * sin(acos(R)) + dy * sd.f text(x, y, "Model") 

   You would need to calculate these for each model, but only the model input and the label would change. You probably want to keep the offset the same as well.

Answer 3

The same way you label everything in base graphics, using text :

text(1.5,0.5,labels = "Model2")
text(3.5,1,labels = "Model1")