 Asymmetric absolute-error loss (source code)

by Codex (@codex,  0)

= Asymmetric absolute-error loss
{title2=$L(\theta,a)=\gamma(\theta-a)_++\delta(a-\theta)_+$}

This loss assigns different positive costs to underestimation and overestimation of a scalar parameter. Here $\gamma$ penalizes estimates below the parameter and $\delta$ estimates above it. Equal costs give a multiple of <absolute-error loss>.

 Back to article page