Weighted samples, kernel density estimators and convergence
Weighted samples, kernel density estimators and convergence
复制标题
DOI：
10.1007/s001810200134
复制
发表时间：
2003-04
影响因子：
3.2
通讯作者：
F. J. G. Gisbert
中科院分区：
文献类型：
--
作者：
F. J. G. Gisbert
本文以几种方式将标准核密度估计推广到加权样本的情况。首先，我考虑通过将估计量定义中的简单和替换为加权和来进行明显的扩展，但我也考虑引入权重的其他替代方案，基于自适应核密度估计，并将权重视为观测信息内容的指标，在这个意义上，作为数据局部密度的信号。所有这些想法都显示在宏观经济收敛问题的背景下使用宾夕法尼亚大学世界表。
This note extends the standard kernel density estimator to the case of weighted samples in several ways. In the first place I consider the obvious extension by substituting the simple sum in the definition of the estimator by a weighted sum, but I also consider other alternatives of introducing weights, based on adaptive kernel density estimators, and consider the weights as indicators of the informational content of the observations and in this sense as signals of the local density of the data. All these ideas are shown using the Penn World Table in the context of the macroeconomic convergence issue.