making sense of models: a parsimonious solution
by Dr. Gary Nan Tie & Dr. Bob Mark models shape decisions We all need to understand how scientists, economists and doctors use models. This is especially true today since models are used to guide our public health policy for pandemics, and we are in the early stages of receiving vaccinations against the SARS-CoV-2 virus1. A subway map, a wind tunnel aircraft mockup, an architect’s virtual preview are simple examples of models. A more sophisticated model consists of mathematical distillations of relevant features of a problem that we want to solve. Mathematical models guide us in understanding and predicting actual behavior for the purpose of making a decision. In particular, a mathematical model is a flexible mechanism which is adaptive and enables inference which: • can explain and predict phenomena • is consistent • can be vetted • is reproducible • can be updated upon
arrival of new information • allows us to extrapolate • articulates our risk performance • can be perturbated and stress tested
• permits ‘What-if?’ investigation • allows comparison with alternative views • guides rational decision making
Mathematical models enable both qualitative and quantitative reasoning. Quantitative reasoning endeavors to numerically estimate a quantity along with how much give and take there is in an estimate. Qualitative reasoning typically describes how networks of relationships interact with one another, inferring possible cascades of consequences that are directionally correct but not necessarily temporally or quantitatively accurate. Qualitative models may arise, for example, because of a lack of data to parameterize a model but their logic is nonetheless useful.
1 / As we go to press, more than 245 million doses have been administered across 107 countries, according to data collected by Bloomberg. The latest rate was roughly 6.79 million doses per day. See https://www.bloomberg.com/graphics/covid-vaccine-tracker-global-distribution/ 2 / See Nan Tie, G. & Mark, B., Sept 2020,’Parsimony;’A Model Risk Paper’, PRMIA Institute
028
Intelligent Risk - April 2021