Identify influential factors

• What: Identify influential factors
• Also known as:
  • Screening
• When:
• Purpose: identify which factors are influential, and should be focussed on, and which can perhaps be ignored within an analysis
• Context:
  • model with factors that can be varied, potentially qualitative
  • Too many factors to be used within the subsequent analysis
  • Leaving out factors does not have significant effects on conclusion
• Step: Application#Analysis
• How:
  • A variety of techniques
    • Quantitative
      • Morris
      • Sobol
      • Random Forest is a classification and regression algorithm based on using many decision trees. It provides a measure of variable importance based on removing the variable as a predictor. Potential biases depending on the dataset structure need to be accounted for.
      • ...
    • Qualitative
      • ...
  • Interpretation varies depending on the factors and output selected
• Why: why this practice is useful in the particular context
• In what situations should you NOT use?
  • May be instead interested in Ranking factors based on sensitivity
• Things that go wrong
• References and Resources
• Optional Comments