If the data to be mapped is nominal in type the different data types may be distinguished by different hues. Such maps visualise differences in type rather than value. It is generally accepted that eight classes are acceptable in these cases. Examples of this type of map might be land-use maps or a geological maps.

Fig 1: a suitable legend for mapping data of different types - nominal data.

This legend design is inappropriate for mapping variations in value or amount of a single variable as there is not a visual heirarchy between the different hues. Mapping variation in value can be acheived with variation of colour saturation; The variation of the amount of white added to the hue. Light shades appear as low values and more saturated shades as higher values. The number of classes that can be visually discerned becomes a factor in choosing the number of classes.

Fig 2: variations in colour saturation limit class number to a maximum of six.

Are your classes appropriate

The differences within each class should be greater than the differences between the classes. By meeting this condition your class boundaries will more closely represent the actual boundaries in the data in the real world \
- homogeneity within a class
- heterogeneity between classes

How do I achieive this?

1. Plot a graphical array of the data. Class boundaries should fall where slope changes are significant.
2. Perform an analysis of variance (ANOVA). The variance of all the classes (mean values) is divided by the variance within the classes. More ¼