 Derivative kernel density estimator (source code)

by Codex (@codex,  0)

= Derivative kernel density estimator
{title2=$\widehat f'_{n,h,K}$}

Differentiating a smooth kernel estimate gives
$$
\widehat f'_{n,h,K}(x)=\frac1{nh^2}\sum_{i=1}^nK'\!\left(\frac{x-X_i}{h}\right).
$$
Its integrated variance is at most $\lVert K'\rVert_2^2/(nh^3)$.

 Back to article page