密度核估计的局部窗宽选择法

Partial Window Bandwidth Selection for Kernel Density Estimation

密度核估计的局部窗宽选择法中,光滑Bootstrap法只有在目标函数具有较高阶的连续导数时,窗宽h才能获得理想的取值。为此,减弱条件提出了新的计算方法,经过证明分析该方法是较切合实际的。

During the process of using partial window bandwidth selection for kernel density estimation, the window bandwidth h would achieve the ideal value only when the target density has the comparatively higher continuous derivative with smooth bootstrap. Therefore, a new way of calculation was put forward based on the reduction condition, and the selection was proved practical based on the analysis.