5.2 Measures of Model Specificity

5.2.1 Generality

The generality of a model is defined by the conditions in which it is applicable. A model is made unambiguously more general by subtracting from the set of its conditions of application. It is made less general (or more specialized) by adding to the set of its conditions of application.

For example, if a model is found sometimes to forecast accurately and other times to forecast inaccurately, the natural procedure is to determine the special conditions in which the model is a useful guide to action. A natural presumption is that additional conditions of application are required. This will require some procedure to discriminate amongst those additional conditions for which the model holds and those in which it does not hold. If such a procedure is successful, a better model will result which differs from its forebear in having a larger number of conditions of application. It will therefore be applied in a more restricted set of cases.

Modelling Learning as Modelling - 23 FEB 98

[Next] [Previous] [Up] [Top] [Contents]

Generated with CERN WebMaker