Short communication: selecting input distributions for use in Monte Carlo simulations

Area of Science:

• Risk Assessment
• Computational Statistics

Background:

• Monte Carlo analysis is widely used for uncertainty quantification in risk assessment.
• The accuracy of Monte Carlo results heavily depends on the choice of input distributions.

Purpose of the Study:

• To present guidelines for selecting appropriate input distributions for Monte Carlo analysis.
• To improve the reliability of uncertainty quantification in risk assessment.

Main Methods:

• Classifying variables based on data range and behavior.
• Employing goodness-of-fit tests when distribution class is unclear.
• Utilizing bootstrapping techniques for sampled distributions.
• Using analogous data for default distributions when site-specific data is absent.

Main Results:

• A systematic approach to selecting input distributions for Monte Carlo simulations.
• Methods to handle situations with unknown or limited data.

Conclusions:

• The proposed guidelines enhance the robustness of Monte Carlo analysis in risk assessment.
• These methods provide practical solutions for selecting input distributions, even with data limitations.