= Exact mean integrated squared error of a kernel density estimator
{title2=$\operatorname{MISE}=R(K_h*f-f)+n^{-1}[R(K_h)-R(K_h*f)]$}
For <independent> observations with common <probability density function>[density] f, the <kernel density estimator> has <mean> $K_h*f$. Its pointwise <variance> is $n^{-1}[(K_h^2*f)-(K_h*f)^2]$. Integrate the <variance> and add integrated squared bias to get the displayed exact <mean integrated squared error>, assuming the <integrals> are finite. Since $R(K_h)=R(K)/h$, this also separates the smoothing bias from the stochastic <variance>.
Back to article page
