Graphics Reference
In-Depth Information
Figure
.
.
MDS (bottom) and parallel coordinate plot (top) of the cars data set. he four cars in the
lower right of the MDS are highlighted and happen to be the most expensive and most powerful cars in
the data set
Limits
6.4.2
Parallel coordinates are oten overrated with respect to the insight they provide into
multivariate features of a data set. Obviously scatterplots are superior for investigat-
ing
-D features, but scatterplot matrices (SPLOMs) need far more space to plot the
same information as PCPs. Even the detection of multivariate outliers is not some-
thing that can usually be directly aided by parallel coordinates. Detecting features
in parallel coordinates that are
not
visibleina
-Dor
-Dplotisrare.Ontheother
hand, parallel coordinate plots are extremely useful for interpreting the findings of
