Generating sets of diverse and plausible scenarios through approximated multivariate normal distributions

Abstract

This article presents a novel and broadly generalizable framework for generating diverse and plausible sets of scenarios. Potential future outcomes are decomposed using a set of uncertainties which are assumed to be multivariate normally distributed, regardless of whether the uncertainties actually present numerically quantifiable phenomena. The optimal scenarios are then chosen along the principal components of the distribution, and the results can be easily interpreted and visualized. Notably, our approach requires a relatively small number of numerical assessments, offering an efficient and practical solution for decision-makers. The framework also provides a testable setting for evaluating its performance and allows users to iteratively improve future-related assumptions and predictions. These findings are relevant for all fields that aim to understand potential future developments, such as, but not limited to, foresight, economics, business strategy and strategic intelligence analysis.