A function dataset for benchmarking in sensitivity analysis

Abstract

In this paper a dataset of functions has been described which includes analytical values of Sobol’ first-order and total-order indices. This unique collection represents a valid benchmark to evaluate sensitivity analysis methodologies and allows the comparison of different technique outcomes. The benchmarking dataset was introduced in Azzini and Rosati following a practice already consolidated in many fields of research such as machine learning or Statistics. The dataset should be considered as an initial proposal open to being easily updated, extended, or modified by new mathematical functions or models in the future.

Keywords

• Function dataset
• Sensitivity analysis benchmarking
• Sobol’ indices
• Analytical sensitivity indices
• Main effect index