Accurate Estimates Without Local Data?

We make no claim that all process models are ‘just right’ and, hence, can be controlled by our methods. Such processmodels can be quite complex
andinclude: discrete-eventmodels (LawandKelton 2000; Kelton et al. 2002); system dynamics models (Abdel-Hamid and Madnick 1991); state-based
models (Harel l990; Akhavi and Wilson 1993; Martin and Raffo 2000); rule-based programs (Mi and Scacchi 1990); or standard programming constructs
such as those used in Little-JIL (Cass et al. 2000; Wise et al. 2000). These rich modeling frameworks allow the representation of detailed insights into an
organization. However, in data-starved domains, the effort required to tune may be non-trivial.

In terms of the Goldilocks principle, we suspect that many process models may not be near the ‘right size’ and will require extensive tuning before they
can be used for decision making. Hence, in domains suffering from a data drought, we would advocate ‘just right’ models like those from USC.