show volume in each node using ctree , plot in R

Question

can any one please show me how to add volume in each of the nodes , instead of the final node volume

t <- ctree(is_return ~ a + b + c) 
plot(t, type="simple")

and my tree would look like

enter image description here

how can I modified that plot where it would show N= on every circle nodes , not only the black or the final node.

Thanks

• agstudy: Can you supply a reproducible example please? And explain what 's N?
• JPC: Oh,N in here is volume , say of the sample, N = 100 , meanings 100 fall into the black square box. but I want to add N in box b and c .. Say if 150 fall into b oval, I want to add N = 150 in there.

Answer 1

enter image description here

The idea is to specify a panel functions for plotting inner nodes.

I generate some data, and the tree

lls <- data.frame(N = gl(3, 50, labels = c("A", "B", "C")), 
 a = rnorm(150) + rep(c(1, 0,150)),
 b = runif(150))
pond= sample(1:5,150,replace=TRUE)
tt <- ctree(formula=N~a+b, data=lls,weights = pond)

The custom inner plot function. I draw a circle where i write the some of weights.

innerWeights <- function(node){
 grid.circle(gp = gpar(fill = "White", col = 1))
 mainlab <- paste( node$psplit$variableName, "\n(n = ")
 mainlab <- paste(mainlab, sum(node$weights),")" , sep = "")
 grid.text(mainlab,gp = gpar(col='red'))
}

I plot the tree

plot(tt, type='simple', inner_panel = innerWeights)

PS: the results depends on a random generated data, so you will not probably get the same plot.

• agstudy: @JPC you are welcome. Next time please try to produce a good reproducible example. when I have time I'll do version with ggplot2 inner_function.
• JPC: I 'm trying to add back node labels ("A", "B", "C" ) into the inner_panel, but didn't have any success. Could you please show me how to ? Thank you.
• agstudy: @JPC One hint , put a browser() in the begining of innerWeights , and see your node argument...
• JPC: thank you ! I need to send you a thank you card! really ! I am very new to R , but the R community has been great to newbies like me.
• agstudy: @JPC Welcome to R but I am a newbie too ! Do you use R for work or study?

Answer 2

When exporting the decision tree, change the dimensions of the image. for a decision tree with 200 terminal nodes the width should be about 30.000.