Bayesian network modeling metrics of performance and uncertainty

Model sensitivity analysis variance reduction ( ) ( ), where ( ) ∑ ( )[ ( )] , ( ) ∑ ( )[ ( )] , ( ) ∑ ( ) , where is the numeric real value of state q, ( ) is the expected real value of Q before applying new findings, ( ) is the expected real value of Q after applying new findings f for variable F, and ( ) is the variance of the real value of Q before any new findings [0,infinity], the greater the value, the more sensitive is a resultant node Q to the node in question F; used with continuous variables Marcot et al. 2006, Marcot 2012