Sobolindices.com: the data-driven online Sobol' indices calculator is now online!

Sensitivity analysis is a very useful tool in virtually any field of engineering and applied sciences.

Performing advanced sensitivity analysis can be quite complex computationally, therefore requiring a large number of model runs and/or specific input parameter exploration strategies. This is in particular true for well established methods like the variance-decomposition-based Sobol’ sensitivity indices[1].

We are therefore happy to launch our new pilot online service www.sobolindices.com to provide an easy[2] way out to anyone who wants to do sensitivity but without the agonizing pain! The entirely data-driven tool capitalizes on the beautiful synergy between sparse polynomial chaos expansions and Sobol’ indices[3], and is powered by UQLab.

Head over to www.sobolindices.com and try it out! If you don’t have a test dataset, you can download this one, which corresponds to a 3-dimensional test case with some interesting features (spoiler alert: it’s the Ishigami function!).

Let us know if you like this pilot project and if you would like to see it grow with more functionality in the future.

  1. Sobol, I.M.: Sensitivity estimates for non linear mathematical models. Math. Model. Comput. Exp. 1 , 407–414 (1993) ↩︎

  2. https://www.sobolindices.com/howto/ ↩︎

  3. Sudret, B.: Global sensitivity analysis using polynomial chaos expansions, Rel. Eng. & Sys. Safety,
    Volume 93, Issue 7, 964–979 (2008) ↩︎