Learn R Programming

DescTools (version 0.99.56)

d.whisky: Classification of Scotch Single Malts

Description

86 malt whiskies are scored between 0-4 for 12 different taste categories including sweetness, smoky, nutty etc. Additionally, coordinates of distilleries allow us to obtain pairwise distance information. Using a combination of these variables it is possible to look for correlations between particular attributes of taste and physical location, for example does a shared local resource have a significant effect on nearby whiskies.
By using correlation analysis it may be possible to provide whisky recommendations based upon an individual's particular preferences. By computing the Pearson correlation coefficient and specifying a threshold value between 0 and 1, we can establish an adjacency matrix where each node is a malt whisky and an edge represents a level of similarity above the threshold.

Usage

data("d.whisky")

Arguments

Format

A data frame with 86 observations on the following 16 variables.

distillery

a character Aberfeldy, Aberlour, AnCnoc, Ardbeg, ...

brand

a grouping factor to separate the better known distilleries (A) from the lesser known ones (B).

region

a factor with levels campbeltown, highland, islands, islay, lowland, speyside.

body

a numeric vector

sweetness

a numeric vector

smoky

a numeric vector

medicinal

a numeric vector

tobacco

a numeric vector

honey

a numeric vector

spicy

a numeric vector

winey

a numeric vector

nutty

a numeric vector

malty

a numeric vector

fruity

a numeric vector

floral

a numeric vector

postcode

a character AB30 1YE, AB35 5TB, ...

latitude

a numeric vector, coordinate pairs of distilleries.

longitude

a numeric vector, coordinate pairs of distilleries.

References

http://www.mathstat.strath.ac.uk/outreach/nessie/index.html

Examples

Run this code
head(d.whisky)

opar <- par(mfrow=c(3,3), cex.main=1.8)
for(i in 1:9)
  PlotPolar(d.whisky[i, 4:15], rlim=4, type="l", col=DescTools::hecru, 
            lwd=2, fill=SetAlpha(DescTools::hecru, 0.4),
            panel.first=PolarGrid(
              ntheta=ncol(d.whisky[i, 2:13]), nr = NA, col="grey",
              lty="dotted", las=1, cex=1.4, alabels=StrCap(colnames(d.whisky[i, 3:14])),
              lblradians=TRUE),
            main=d.whisky[i, "distillery"])


par(mfrow=c(3,3), cex.main=1.8, xpd=NA)
id <- d.whisky$distillery %in% c("Ardbeg","Caol Ila","Cragganmore","Lagavulin","Laphroig",
                                   "Macallan","Mortlach","Talisker","Tobermory")
PlotFaces(d.whisky[id, 4:15], nr=3, nc=3, col=hecru, scale=TRUE, fill=TRUE,
          labels=d.whisky$distillery[id])

par(opar)

Run the code above in your browser using DataLab