The VC dimension and ecological modelling

In this article, a measure of model complexity known as the VC dimension is presented and some mathematical theory surrounding the derivation and use of this metric is summarized. The metric allows in some instances, for a modeller to control the amount of error that is returned from a modelling system. Moreover, it allows modellers to state upper bounds on the amount of error that the modelling system will return on all future, as yet unseen and uncollected datasets. Some computational experiments utilizing this theory in a series of models predicting the spatial distribution of plant species on the east coast of Australia are presented and their results are discussed.