Decision Tree: What do all the colours mean?


In all the views the colour coding used is as follows:

  • Red: Relates to people in the analysis selection

  • Blue: Relates to people not in the analysis selection

In the graphical views, the intensity of the colour relates to the gain of the node, relative to the root node.  The Gain is the ratio of the Analysis % in the node, to the Analysis % in the root.  Nodes are coloured:

  • Red if they have a gain greater than 1

    The more intense the red, the higher the proportion of people in the analysis selection.

  • Blue if they have a gain less than 1

    The more intense the blue, the lower the proportion of people in the analysis selection.

  • Grey if they have a gain around 1 (i.e. are in line with the root node).

Note, if the root has a high Analysis % (e.g. 70%), colours will tend to be grey or muted since gain figures will be closer to 1.

The same colour coding is used in the Next Splits panel, but is applied at 2 levels: the individual category and the branch.  For more details see What can I learn from the Next Splits panel?

The colours used can be changed using the Options menu.