MODEL TRANSFERABILITY
| Collection | International Encyclopedia of Systems and Cybernetics |
|---|---|
| Year | 2004 |
| Vol. (num.) | 2(2) |
| ID | ◀ 2164 ▶ |
| Object type | Epistemology, ontology or semantics, Methodology or model |
The possibility to generalize a particular model in order to use it in a variety of cases.
P.M. ALLEN et al. propose the following — quite systemic — way to construct general models: “Instead of building a model of a particular system (New York, the insurance market in Belgium, the Aleutian ecosystem) in terms of the salient features of the moment, it may be more informative to study generic models (urban systems, markets, interacting populations…) and the ways that their constituent elements cluster, grow, and structure according to the series of ”folds“ that their interactive mechanisms impart to them, and the attributes which emerge for different branches of solutions.
- “The study of such evolutionary trees may then show us that New York, Brussels, and other cities correspond to different branches of solution of similar models with different parameter values and different histories. Only a ”transferable“ model of this kind can be used predictively on a particular case, since a model built in terms of a particular problem, and calibrated on it, is a purely descriptive construct incapable of anticipating change. So,… a good model must be a metamodel!” (1985, p.76).
… Or at least derived of such a general frame in accordance with the specific situation under study.
And the best metamodels are those which retrace analogies, homomorphies and isomorphies.