Sensitivity analysis in logic-based models

摘要

An important feature of the causal models used in most DSS is that they allow their users to perform sensitivity analyses. Most of these analyses are performed on continuous variables, which are the principal types of variables found in causal models. However, some of the models derived from artificial intelligence that are increasingly being used in DSS contain logic (or Boolean) variables. In this paper we exploit the notion of a ‘Boolean derivative’ developed for circuit design and apply it to the calculation of sensitivity measures in logic-based models.