Loss function

statistical decision theory

$Loss(\theta,d)$

the best decision therefore is the value which minimizes the expected loss

$E[Loss(d,\theta)|x]$