o MANET


Mosaicplot


First step: a barchart
A mosaicplot is a multidimensional barchart. A barchart divides the data-set into categories. If you want to split the data by more than one classification, you need a mosaicplot.

Two dimensions
Each bin in a barchart is made a barchart itself, but this time vertically. Turn your head 90 degrees and you can see 4 barcharts one on top of the other.

Try how far it gets you
The next division is made horizontally again, and so on.
In the end you get little boxes (well, some are large) for every combination of attributes. Currently up to 256 are permitted (you are free to have up to 8 variables each with only two categories or smaller numbers of variables with more categories). MANET could be extended to cope with larger numbers and this would be practicable if cinema-size screens become commonplace.


What else?
It is not necessary that all variables have the same number of categories; here is an example with one variable falling into two and one into three. As you can see the sequence of the variables changes the look of the plot even with only these two variables.
This changing can be done via a parameter-list or directly with the arrow-keys. The parameter-list offers the option of reducing the distances between the boxes to zero. This version of the display is called mondrian.

Why mondrian? Have a look at one of Mondrian's typical pictures and guess.

An implementation of Mosaic-Plots for S-Plus (v0.9)
Further options:



back to MANET