Graphics Reference
In-Depth Information
Figure
.
.
Fluctuation diagram for a correlation matrix. he correlations between monthly average
temperatures at different locations are plotted for three years. Seasonal trends yield the dominant
structure in the picture
Ash,Fat,Lactose,Protein,andWater.Weusethepercentagesoftheingredients ofthe
milks for each animal as weights in the mosaicplot. Figure
.
shows two weighted
mosaics given in fluctuation mode for the Mammals Milk data. he cases are listed
inalphabeticorder.heletsideshowstherawdata;wecanseethatallofthemilks
mostly consist of water. DOLPHIN and SEAL show up as outliers, since they have
the lowest percentages of Water, and highest percentages of Fat.
he different ingredients are allocated rather unevenly, so the use of standardized
valuesherehelpstodetectmoredetailedpatternsbetweeneachpairofmammals.he
right side of Fig.
.
shows standardized values (via X
σ
X
)usedasweights.
Again, it is obvious that DOLPHIN and SEAL behave similarly, but now we can also
see that DEER and REINDEER also are very similar. Surprisingly, BISON and BUF-
FALO seem to be very different.
he use of clustering techniques like Complete Linkage yields results like those
shown in Fig.
.
. he rows of the data table are sorted according to the hierarchi-
cal clustering scheme. Now we can check the results from the clustering algorithm
graphically by comparing the rows of the mosaicplot. his enables us to choose the
number of clustersby graphical means, which, of course, is a sot and very subjective
criterion, but one that enables us to respond quickly to different situations. In this
X
