DATA MINING Desktop Survival Guide by Graham Williams
Desktop Survival Project Home List of Figures List of Tables Data Mining Data Mining Data Mining with Rattle Introduction Data Transform Explore A Model Building Framework Unsupervised Modelling Two Class Models Multi Class Models Regression Models Text Mining Evaluation and Deployment Moving into R Troubleshooting R for the Data Miner R Data Graphics in R Understanding Data Preparing Data Building Models Evaluating Models Algorithms Apriori Bagging Bayes Classifier Boosting Cluster Analysis Conditional Trees Hierarchical Clustering K-Means K-Nearest Neighbours Linear Models Logistic Regression Neural Networks Support Vector Machines Text Mining Open Products AlphaMiner Borgelt Data Mining Suite KNime R Rattle Weka Closed Products C4.5 Clementine Equbits Foresight GhostMiner InductionEngine ODM Enterprise Miner Statistica Data Miner TreeNet Virtual Predict Appendicies Glossary Bibliography Index

Tuning a Boxplot

Here we illustrate how we can refine exactly what we want to draw in the box plot. Three boxplots are produced on the single plot using par to set mfrow to one row and three columns. We then collect the output from the boxplot function which we might look at to determine inforamtion about what is being plotted. In this case we might decide to set the limits of the boxplot to be 0 and 5.2, and we note the other statistics in the stats attribute of the output.

> boxplot.info
$stats
      [,1]
[1,] 0.340
[2,] 1.200
[3,] 2.135
[4,] 2.880
[5,] 5.080

$n
[1] 178

$conf
         [,1]
[1,] 1.936044
[2,] 2.333956

$out
numeric(0)

$group
numeric(0)

$names
[1] "1"

The bxp function (used internally by boxplot) is used to plot the boxplot.

We now modify the boxplot information (the stats attribute) to use 10% and 90% deciles (obtained using quantile) instead of the default 0% and 100% deciles.

> deciles
   0%   10%   20%   30%   40%   50%   60%   70%   80%   90%  100% 
0.340 0.607 0.872 1.324 1.738 2.135 2.460 2.689 2.980 3.233 5.080

This generates the second boxplot.

Finally, a completely different boxplot showing the mean +/- one and two standard deviations, is produced. The structure used by bxp for recording the information to be plotted is built from scratch, in particular the 5 stats used to specify where the limits go.

[width=0.8]rplot-wine-boxplot-tuning

oldpar <- par(mfrow=c(1,3))

x <- wine$Flavanoids
boxplot.info <- boxplot(x, plot=FALSE)
bxp(boxplot.info, main="Normal boxplot", ylim=c(0,5.2))

deciles <- quantile(x, probs=seq(0,1,0.1))
boxplot.info$stats[1] <- deciles["10%"]
boxplot.info$stats[5] <- deciles["90%"]
bxp(boxplot.info, main="10%/90% whiskers", ylim=c(0,5.2))

boxplot.limits <- as.matrix(c(mean(x) - 2*sd(x),
                            mean(x) - sd(x),
                            mean(x),
                            mean(x) + sd(x),
                            mean(x) + 2*sd(x)))
boxplot.meansd <- list(stats = boxplot.limits,
                       n = length(x),
                       conf = NULL,
                       out = numeric(0))
bxp(boxplot.meansd, main=expression("mean" %+-% "1 and 2 SDs"), ylim=c(0,5.2))

par(oldpar)

http://rattle.togaware.com/code/rplot-wine-boxplot-tuning.R

Support further development through the purchase of the PDF version of the book.
Brought to you by Togaware.