Dimensional Stacking
This technique projects high-dimensional data by embedding
dimensions within other dimensions. It starts by
discretizing the ranges of each dimension. Each dimension
is then assigned an orientation and an ordering. The
dimensions with 2 lowest ordering are used to divide a
virtual screen into sections, with the cardinality to
determine how many sections horizontally and vertically
will be generated. Each section is then used to define the
virtual screen for the next 2 dimensions, again using the
cardinality to determine how to break up the virtual screen.
This is repeated until all dimensions have been embedded and
the data point can be mapped to its screen location.
In this approach, the clusterings tend to appear as
repeating patterns in the inner-most dimensions.
Transitions in the pattern indicate clusters which are
shifting along one or more dimension. This method is most
useful for dense data sets, as then more of the buckets get filled.
