Kernel estimation of conditional density with truncated, censored and dependent data
Kernel estimation of conditional density with truncated, censored and dependent data
复制标题
使用截断、删失和相关数据对条件密度进行核估计
DOI：
10.1016/j.jmva.2013.05.009
复制
发表时间：
2013
影响因子：
1.6
通讯作者：
Liu, Ai-Ai
中科院分区：
文献类型：
--
作者：
Liang, Han-Ying;Liu, Ai-Ai
本文基于条件分布函数的广义乘积极限估计，定义了一类左截尾右删失模型的条件密度的核估计。在协变量为平稳α混合序列的观测值的情况下，我们得到了估计的渐近正态性质和Berry-Esseen界。此外，还考虑了估计量的一致收敛速度。通过仿真研究了估计器的有限样本行为。
In this paper we define a kernel estimator of the conditional density for a left-truncated and right-censored model based on the generalized product-limit estimator of the conditional distributed function. Under the observations with multivariate covariates form a stationary α-mixing sequence, we derive the asymptotic normality as well as a Berry–Esseen type bound for the proposed estimator. Also, the uniform convergence with rates for the estimator is considered. Finite sample behavior of the estimator is investigated via simulations too.