Return on modeling effort explained

Return on modelling effort (ROME) is the benefit resulting from a (supplementary) effort to create and / or improve a model.^{[1] [2]}

Purpose

In engineering, modelling always serves a particular goal. For example, the lightning protection of aircraft can be modelled as an electrical circuit, in order to predict whether the protection will still work in 30 years, given the ageing of its electrical components. More and more effort can be put in making this model predict reality perfectly. However, this perfection comes at a price: researchers invest time and money in improving the model. As a Return on investment (ROI), the ROME is a metric for the use of further modelling. It may therefore serve as a 'stopping criterion'.

Typically, researchers will pull towards continuing modelling, while management will pull towards stopping modelling. Being explicit about the cost and benefits of continued modeling may help to make informed decisions that are understood by both sides. Continuous communication between model developers and model users increases the probability of models being actually put to profit.^[3]

Domains

ROME is a metric, which can be evaluated wherever modelling is performed with a quantifiable goal. Examples include:

• Modeling the impact of federal policy on social problems.
• Modeling a marketing mix to statistically correlate a number of inputs (or independent variables) – such as a marketing campaign – to outcomes (or dependent variables) – such as sales or profits.^[4]
• Modeling the links between enterprise actors to make an informed choice on splitting organizations.^{[5] [6]}
• Modeling the coupling of an electromagnetic interference to a PCB to reduce its susceptibility by improving the routing of traces.^[7]

Research

The initiative "Models at Work" studies the creation, management and use of domain models in scientific and industrial practice, aiming at a diversity of goals, varying from (as truthful as possible) representation of the conceptual structure of the domain that is modelled, via animation, simulation, execution and gamification, until automated (logic-based) reasoning.^[8]

References

1. Book: 10.1007/978-3-540-85232-2. Chapter 4 of Enterprise Architecture - Creating Value by Informed Governance. M.. Op 't Land. H. A.. Proper. M.. Waage. J.. Cloo. C.. Steghuis. Springer, Heidelberg, Germany. 2008. 978-3-540-85231-5.
2. Web site: Models that matter; Return on Modelling Effort. erikproper.blogspot.fr. Proper. Erik. 2009. 2016-10-06.
3. Fromm . Gary . Gass . Saul I. . Review of the non-DoD modeling effort . Utility and Use of Large-scale Mathematical Models: Proceedings of a Workshop . April 28–29, 1977 . 534 of NBS special publication . 15–21 . Department of Commerce, National Bureau of Standards, 1979 . Gaithersburg, Maryland.
4. Web site: Marketing Mix Modeling. Thomas. Jerry W.. 2006. 2016-10-06.
5. Book: Op 't Land, Martin. Instrument for fast and effective splitting of organizations. Delft University of Technology. 2008. 978-90-71382-37-6.
6. Book: Op 't Land, Martin. Applying Architecture and Ontology to the Splitting and Allying of Enterprises. Delft University of Technology. 2008. 978-90-71382-32-1.
7. Op 't Land. Sjoerd. Mandić. Tvrtko. Ramdani. Mohamed. Barić. Adrijan. Perdriau. Richard. Nauwelaars. Bart. 2. Comparison of Field-To-Line Coupling Models: Coupled Transmission Lines Model versus Single-cell Corrected Taylor Model. Proceedings of the 2013 International Symposium on Electromagnetic Compatibility (EMC EUROPE 2013). September 2013 . 276–281 .
8. Web site: Models at Work – Case reports from industry and science, on the creation and use of domain models. Proper. Erik. 2022-02-22.