Binding of Predictive Models

Decision models are rule/expression based: the inputs and output are usually structured data, which can be quantitative or categorical.

Predictive models (aka “AI models” in the narrow sense) are more mathematical in nature (thus requiring different engines), and process inputs “features” that are usually quantitative only, and may result from a ‘mapping’ (see for example the Transformations defined as part of the PMML standard http://dmg.org/pmml/v4-4-1/Transformations.html)