The predictive model, by its very nature, is used toforecast the future. Predictive models are based on theassumption that the situation is an established ratherthan an emergent one. In the case of an establishedsituation, the relevant environmental conditionsare given and predictable, whereas in an emergentsituation some of these conditions do not prevail andtherefore the data generated is often qualitative ratherthan quantitative. This is a significant point, sincean important distinction between predictive modelsand evaluative models is the type of data generatedthrough their use. The function of predictive models ispredominantly to produce quantitative data, whereasthe use of evaluative models generally provides dataof a qualitative nature. This is found to be particularlyrelevant to the case of architecture, a discipline inwhich predictive models enable the designer to measurethe effects of changing design variables rather thanperceive them, since there can be no ‘difference whichdoesn’t make a difference’. Another distinction betweenpredictive and evaluative architectural models is seenin the stage at which they are used during the design  process. As described above, an evaluative model is usedwhen many of the design variables have been decidedin order that their effects can be perceived. By contrast,a predictive model is usually used earlier in the designprocess so that such design variables can be determinedby its use. Over the last two decades, computer softwarehas enabled increasingly complex information to bedigitally simulated, whether energy flows, climateconditions or the implications of inhabitation. Withplatforms such as EcoTect and various plugins for Rhinosuch as Kangaroo, Pachyderm, etc., the relationshipbetween environmental conditions and buildingperformance may be modelled very precisely. However,the ability of physical models to be able to integrate withsuch data and represent it has led to new hybrid typesof models, some examples of which are illustrated inthis subsection.