[R] ctree
Achim Zeileis
Achim.Zeileis at uibk.ac.at
Wed Nov 14 19:47:18 CET 2012
On Wed, 14 Nov 2012, mia88 wrote:
> Hello, I plotted a nice tree with "ctree" . It shows 3 nodes with the
> prediction of my 2 groups. (see picture) Unfortunately I need a larger
> scale to read the exact prediction of my groups to get the specificity
> and sensitivity. I tried to change the scale with "axis" but it didn't
> work, my guess because it's not a normal graph with x and y axis. Has
> someone an idea how to change the scales in the nodes of my tree?
I'm not exactly sure how you would like to improve the visualization. I
could think of two options: (1) Use a y-axis labeling with only the
predicted probability (rounded to some precision). (2) Add some label,
e.g., at the bottom of the bars, with the predicted probability.
While both options are in principle possible, neither is available out of
the box. It requires some programming using the "grid" package in which
the party plots are created. However, if you start out from the
node_barplot() function provided in the package, not a lot of programming
is necessary. As an example, I have created a function node_barplot2()
with a modified grid.yaxis() call to implement option (1) outlined above.
See the code below. To try it out, please source the entire function
_including_ the class() assignment at the end. And then you can do:
## package and tree for Pima Indians Diabetes data
library("party")
data("PimaIndiansDiabetes", package = "mlbench")
ct <- ctree(diabetes ~ ., data = PimaIndiansDiabetes)
## visualizations: default and alternative y-axis and more spaces
plot(ct)
plot(ct, terminal_panel = node_barplot2)
plot(ct, terminal_panel = node_barplot2,
tp_args = list(ylines = c(2, 4)))
Hope that helps,
Z
node_barplot2 <- function(ctreeobj,
col = "black",
fill = NULL,
beside = NULL,
ymax = NULL,
ylines = NULL,
widths = 1,
gap = NULL,
reverse = NULL,
id = TRUE)
{
getMaxPred <- function(x) {
mp <- max(x$prediction)
mpl <- ifelse(x$terminal, 0, getMaxPred(x$left))
mpr <- ifelse(x$terminal, 0, getMaxPred(x$right))
return(max(c(mp, mpl, mpr)))
}
y <- response(ctreeobj)[[1]]
if(is.factor(y) || class(y) == "was_ordered") {
ylevels <- levels(y)
if(is.null(beside)) beside <- if(length(ylevels) < 3) FALSE else TRUE
if(is.null(ymax)) ymax <- if(beside) 1.1 else 1
if(is.null(gap)) gap <- if(beside) 0.1 else 0
} else {
if(is.null(beside)) beside <- FALSE
if(is.null(ymax)) ymax <- getMaxPred(ctreeobj at tree) * 1.1
ylevels <- seq(along = ctreeobj at tree$prediction)
if(length(ylevels) < 2) ylevels <- ""
if(is.null(gap)) gap <- 1
}
if(is.null(reverse)) reverse <- !beside
if(is.null(fill)) fill <- gray.colors(length(ylevels))
if(is.null(ylines)) ylines <- if(beside) c(3, 4) else c(1.5, 2.5)
### panel function for barplots in nodes
rval <- function(node) {
## parameter setup
pred <- node$prediction
if(reverse) {
pred <- rev(pred)
ylevels <- rev(ylevels)
}
np <- length(pred)
nc <- if(beside) np else 1
fill <- rep(fill, length.out = np)
widths <- rep(widths, length.out = nc)
col <- rep(col, length.out = nc)
ylines <- rep(ylines, length.out = 2)
gap <- gap * sum(widths)
yscale <- c(0, ymax)
xscale <- c(0, sum(widths) + (nc+1)*gap)
top_vp <- viewport(layout = grid.layout(nrow = 2, ncol = 3,
widths = unit(c(ylines[1], 1, ylines[2]), c("lines", "null", "lines")),
heights = unit(c(1, 1), c("lines", "null"))),
width = unit(1, "npc"),
height = unit(1, "npc") - unit(2, "lines"),
name = paste("node_barplot", node$nodeID, sep = ""))
pushViewport(top_vp)
grid.rect(gp = gpar(fill = "white", col = 0))
## main title
top <- viewport(layout.pos.col=2, layout.pos.row=1)
pushViewport(top)
mainlab <- paste(ifelse(id, paste("Node", node$nodeID, "(n = "), "n = "),
sum(node$weights), ifelse(id, ")", ""), sep = "")
grid.text(mainlab)
popViewport()
plot <- viewport(layout.pos.col=2, layout.pos.row=2,
xscale=xscale, yscale=yscale,
name = paste("node_barplot", node$nodeID, "plot",
sep = ""))
pushViewport(plot)
if(beside) {
xcenter <- cumsum(widths+gap) - widths/2
for (i in 1:np) {
grid.rect(x = xcenter[i], y = 0, height = pred[i],
width = widths[i],
just = c("center", "bottom"), default.units = "native",
gp = gpar(col = col[i], fill = fill[i]))
}
if(length(xcenter) > 1) grid.xaxis(at = xcenter, label = FALSE)
grid.text(ylevels, x = xcenter, y = unit(-1, "lines"),
just = c("center", "top"),
default.units = "native", check.overlap = TRUE)
grid.yaxis()
} else {
ycenter <- cumsum(pred) - pred
for (i in 1:np) {
grid.rect(x = xscale[2]/2, y = ycenter[i], height = min(pred[i], ymax - ycenter[i]),
width = widths[1],
just = c("center", "bottom"), default.units = "native",
gp = gpar(col = col[i], fill = fill[i]))
}
if(np > 1) {
grid.text(ylevels[1], x = unit(-1, "lines"), y = 0,
just = c("left", "center"), rot = 90,
default.units = "native", check.overlap = TRUE)
grid.text(ylevels[np], x = unit(-1, "lines"), y = ymax,
just = c("right", "center"), rot = 90,
default.units = "native", check.overlap = TRUE)
}
if(np > 2) {
grid.text(ylevels[-c(1,np)], x = unit(-1, "lines"), y = ycenter[-c(1,np)],
just = "center", rot = 90,
default.units = "native", check.overlap = TRUE)
}
grid.yaxis(at = round(1 - pred[i], digits = 3), main = FALSE)
}
grid.rect(gp = gpar(fill = "transparent"))
upViewport(2)
}
return(rval)
}
class(node_barplot2) <- "grapcon_generator"
> Thank you very much! Mia
> <http://r.789695.n4.nabble.com/file/n4649478/nodes.png>
>
>
>
> --
> View this message in context: http://r.789695.n4.nabble.com/ctree-tp4649478.html
> Sent from the R help mailing list archive at Nabble.com.
>
> ______________________________________________
> 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.
>
More information about the R-help
mailing list