Better decision tree graphics for rpart via party and partykit

May 29, 2012
By

(This article was first published on Super Nerdy Cool » R, and kindly contributed to R-bloggers)

I’ve been using Graphviz to create better decision tree graphics “by hand” for rpart objects created in R (final tree). I stumbled on this post that shows how one could convert an rpart object to a party project via the as.party function in partykit to utilize the plot functions in party. It looks quite nice.

I might have to do additional hacking as I like to display the node size and percentage of success in every node. For example, in rpart, I do something like

## rpartObj created from rpart
textRpartCustom <- 
{
    nclass <- (ncol(yval) - 1L)/2
    group <- yval[, 1L]
    counts <- yval[, 1L + (1L:nclass)]
    if (!is.null(ylevel)) 
        group <- ylevel[group]
    temp1 <- rpart:::formatg(counts, digits)
    if (nclass > 1) {
        ## temp1 <- apply(matrix(temp1, ncol = nclass), 1, paste, 
        ##     collapse = "/")
      temp1 <- matrix(as.numeric(temp1), ncol=nclass)
      ##temp1 <- paste("p=", round(temp1[, 2] / apply(temp1, 1, sum)*100, 1), "%", "; N=", apply(temp1, 1, sum), sep="")
      temp1 <- paste("", round(temp1[, 2] / apply(temp1, 1, sum)*100, 1), "%", "; ", apply(temp1, 1, sum), sep="") 
    }
    if (use.n) {
        out <- paste(format(group, justify = "left"), "\n", temp1, 
            sep = "")
    }
    else {
        out <- format(group, justify = "left")
    }
    return(out)
}

rpartObj$functions$text <- textRpartCustom
plot(rpartObj)
text(rpartObj)

to get these information to display for a classification fit.

To leave a comment for the author, please follow the link and comment on his blog: Super Nerdy Cool » R.

R-bloggers.com offers daily e-mail updates about R news and tutorials on topics such as: visualization (ggplot2, Boxplots, maps, animation), programming (RStudio, Sweave, LaTeX, SQL, Eclipse, git, hadoop, Web Scraping) statistics (regression, PCA, time series, trading) and more...



If you got this far, why not subscribe for updates from the site? Choose your flavor: e-mail, twitter, RSS, or facebook...

Comments are closed.