Skip to content

summary.md

Latest commit

 

History

History
83 lines (52 loc) · 4.33 KB

summary.md

File metadata and controls

83 lines (52 loc) · 4.33 KB

Extended Summary

Chapter 1. Methodological Landscape of Sensitivity Analysis and the Place of SimDec

Authors: Mariia Kozlova, Samuele Lo Piano, and Julian Scott Yeomans
Source: Kozlova, M., & Yeomans, J. S. (Eds.). (2024). Sensitivity Analysis for Business, Technology, and Policymaking: Made Easy with Simulation Decomposition (SimDec). Taylor & Francis. https://doi.org/10.4324/9781003453789
License: CC BY-NC-ND 4.0

📖 Read full Chapter 1: Ch1.pdf

🎥 Watch the presentation of this chapter: YouTube video

1 · Purpose of the chapter

This chapter explores the methodological foundations of sensitivity analysis (SA) and situates SimDec within this landscape. It evaluates the entire process of sensitivity analysis, from input selection and uncertainty modeling to visualization, and highlights the limitations of current practices in both academia and industry. The chapter establishes the need for a more actionable, interpretable, and complete sensitivity method—fulfilled by the Simulation Decomposition (SimDec) approach.

2 · Full sensitivity analysis process

The authors outline a five-phase sensitivity analysis process:

  1. Input/output selection
  2. Modeling uncertainty
  3. Sampling from distributions
  4. Computing sensitivity indices
  5. Visualization and interpretation

Despite its importance, this full pipeline is rarely followed in practice. Instead, most practitioners stop at Monte Carlo simulations or use inadequate one-at-a-time (OAT) methods.

3 · Critical review of methods

The chapter critiques popular SA techniques:

  • OAT methods (e.g., spider and tornado charts) oversimplify nonlinear models and ignore interactions.
  • Monte Carlo simulations reveal uncertainty but fail to identify drivers.
  • Global Sensitivity Analysis (GSA) (e.g., Sobol’ indices, Shapley values) quantifies effect strength but misses effect shape and dependencies.
  • Visualization remains neglected or insufficient, limiting interpretability and decision-making value.

4 · The SimDec solution

SimDec addresses the above gaps by:

  • Using simple binning to compute both main and interaction effects, even for correlated inputs.
  • Producing stacked, color-coded histograms that expose the shape and source of variability in model output.
  • Enabling intuitive interpretation without compromising analytical rigor.

SimDec integrates the entire SA pipeline and emphasizes visual storytelling of model behavior. It’s especially suitable for decision contexts where stakeholder communication matters.

🔗 Explore the open-source SimDec code on GitHub: Simulation-Decomposition
🧪 Try the interactive dashboard: simdec.io
🎥 Learn how to use the dashboard: Video tutorial
🎥 Learn how to interpret results: Reading SimDec results

5 · Sensitivity analysis in practice

Despite its importance, global sensitivity analysis is used in fewer than 0.03% of computational modeling studies, with business and industry often relying on OAT methods or no SA at all. Adoption barriers include complexity, lack of standardization, and visualization gaps.

The chapter argues that SimDec offers a breakthrough, making comprehensive, interpretable SA feasible and appealing even to non-experts.

6 · Key take-aways

  1. Most SA methods are incomplete or misleading for real decision-making.
  2. SimDec uniquely integrates quantification and visualization, bridging the gap.
  3. Visual, interpretable SA is essential for communication and trust in model results.
  4. Adoption of robust SA methods in practice remains low, but SimDec can help change that.

Attribution

Based on Chapter 1 of Sensitivity Analysis for Business, Technology, and Policymaking
© Mariia Kozlova, Samuele Lo Piano, and Julian Scott Yeomans, 2024 — CC BY-NC-ND 4.0.
This summary is an independent derivative work created for educational and indexing purposes, not affiliated with the original publisher.