Skip to content

summary.md

Latest commit

 

History

History
113 lines (75 loc) · 4.93 KB

summary.md

File metadata and controls

113 lines (75 loc) · 4.93 KB

Extended Summary

Chapter 2. SimDec Algorithm and Guidelines for Its Usage and Interpretation

Authors: Mariia Kozlova, Robert J. Moss, Pamphile Roy, Abid Alam, 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 2: Ch2.pdf

🎥 Watch the introduction to this chapter: YouTube video

🎥 SimDec dashboard tutorial: Watch on YouTube
🎥 How to interpret SimDec results: Watch on YouTube
🧪 Open-source packages: Simulation-Decomposition GitHub
📊 Online interface: simdec.io

Purpose of the Chapter

This chapter introduces the SimDec algorithm, the computational foundation behind the visual decomposition of model outputs. It explains how SimDec:

  • Uses sensitivity indices to identify influential input variables
  • Constructs decomposed output visualizations
  • Helps analysts understand variable influence, interactions, and correlations in computational models
  • Supports transparent, decision-relevant simulations

SimDec Algorithm: Key Steps

The SimDec procedure involves:

  1. Computing global sensitivity indices using a binning-based method
  2. Selecting input variables that explain output variability (threshold-based)
  3. Dividing variables into states (2 or 3, depending on context)
  4. Forming scenarios as all combinations of states
  5. Mapping each simulation output to its corresponding scenario
  6. Visualizing output distributions using color-coded stacked histograms or box plots

📌 See Figure 2.2 on page 6 of the chapter for a step-by-step overview of the decomposition process.

Understanding Sensitivity Indices

SimDec employs three types of sensitivity metrics:

  • First-order effects – measure individual variable contributions
  • Second-order effects – measure interactions or redundancy (overlap/correlation)
  • Combined indices – sum of both, used for variable selection

These indices are computed automatically and guide the decomposition.

How to Read SimDec Visualizations

SimDec’s decomposed histograms allow analysts to:

  • See how inputs shift the output distribution
  • Assess the strength of influence by the alignment (or separation) of colored distributions
  • Spot joint effects (interactions or correlations) by analyzing overlapping or diverging patterns
  • Interpret heterogeneous behavior in complex models with minimal math

Available Packages and Tools

SimDec is available as open-source code in multiple environments:

Platform Features
Python Full functionality with PyPI install, optional dashboard (see example)
R Auto and custom decomposition, example dataset, ggplot2-based visuals
Julia Visualization support with Pluto.jl notebook
Matlab Manual install, supports both sensitivity index calc and plotting
Excel VBA-based Monte Carlo + decomposition interface (no sensitivity indices)

📦 All code: GitHub – Simulation-Decomposition
🖥️ Dashboard: simdec.io

Practical Implementation Guidance

The chapter includes detailed advice on:

  • ✅ How to select input variables (auto vs. manual)
  • ✅ How to form variable states and scenarios
  • ✅ Sampling strategies (random vs. quasi-random)
  • ✅ Recommended sample sizes (≥1,000 for stable indices)
  • ✅ When to use box plots instead of histograms
  • ✅ Limitations and noise handling in small datasets

These guidelines are essential for applying SimDec effectively in real-world modeling.

Key Takeaways

  • SimDec transforms sensitivity analysis into interpretable, visual insights
  • It is fully automated, works with any simulation or dataset with input/output columns
  • Open-source SimDec tools exist across major platforms — with or without coding
  • The algorithm emphasizes actionability, not just diagnostics

Attribution

Based on Chapter 2 of Sensitivity Analysis for Business, Technology, and Policymaking: Made Easy with Simulation Decomposition (SimDec)
© Mariia Kozlova, Robert J. Moss, Pamphile Roy, Abid Alam, and Julian Scott Yeomans, 2024 — CC BY-NC-ND 4.0
This summary is an independent derivative created for educational and indexing purposes, not affiliated with the original publisher.