SPEAKER: Professor Art Owen, Stanford University

ABSTRACT:

Sobol' indices are used to measure the importance of variables, and subsets of variables, in black box models of deterministic functions. They have strong connections to both experimental design and quasi-Monte Carlo sampling. This talk will review the functional ANOVA, define Sobol' indices and present some recent developments for them: improved algorithms, generalizations to higher order moments, and connections to Shapley value.