= Plug-in bandwidth selection
{title2=$\widehat h=[R(K)/(\mu_2(K)^2n\widehat R(f''))]^{1/5}$}
Estimate the unknown curvature in the AMISE minimizer and substitute it into that formula. A consistent positive curvature estimate makes the selected bandwidth ratio-consistent. Pilot smoothing is needed because differentiating an undersmoothed <probability density function>[density] <estimator> magnifies <variance>. A normal-reference pilot can initialize an estimate of a higher <derivative> functional; the final density-curvature estimate is computed from the data rather than assuming the <probability density function>[density] is normal.
Back to article page
