Parallel coordinates

Parallel co-ordinate representation of multi-variate data sets provides a method of visualizing hundreds of variable interactions on one 2D graph.

The parallel coordinate method is based upon replacing the fundamental assumption of orthogonality between dimensions that has restricted previous co-ordinate systems to visualisation in a maximum of three dimensions. If dimensions are instead represented as parallel to each other then the limitation on their number (almost) disappears.

A n-dimensional point can be transformed to parallel co-ordinates. Each axis has its own scale, and the datapoint is described as a polygonal line connecting the individual axis values. To recap: a point in cartesian coordinates is represented by a line in parallel coordinates.

Adding more dimensions to the parallel coordinate plot simply involves adding more axes to the right of the plot and extending the polygonal line to join up with the new points.