ORAHS2025
Abstract Submission

110. Comparative analysis of parametric and nonparametric probability distributions for uncertainty quantification in simulation studies

Invited abstract in session MC-1: Simulation 1, stream Sessions.

Monday, 11:00-12:30
Room: NTNU, Realfagbygget R5

Authors (first author is the speaker)

1. Sara Garber
Statistics and Data Science, University of Augsburg

Abstract

When modeling and analyzing parameters subject to uncertainty, e.g., in Monte Carlo simulation studies, the triangular distribution is frequently used due to its intuitive interpretability, making it particularly accessible to practitioners. However, its simplistic assumptions can introduce bias, potentially affecting the reliability of modeling results in various applications, such as decision support systems. To address this issue, our paper presents a comparative analysis of various parametric probability distributions, including the triangular, beta, and two-sided power distribution, as well as nonparametric methods such as kernel density estimation with various kernel functions. By combining simulation-based approaches with analytical methods, we systematically assess their suitability for modeling, considering key properties of real-world data. Our empirical analysis focuses on a medical use case. This work provides insights into the practical implications of distributional choices in both simulation and analytical modeling and is aimed at statisticians and quantitative researchers involved in the design and methodological refinement of (medical) modeling and simulation studies.

Keywords

Status: accepted

Back to the list of papers