摘要
本文利用核回归估计方法对二元极值Copula函数的相关函数进行估计。构建了相关函数的N-W核回归估计。在选择最优带宽的前提下,通过数值模拟对比了N-W核回归估计与OLS估计。数值模拟的结果显示N-W核回归估计在一定情况下较之于OLS估计更具有稳定性,是一种相对较优的相关函数非参数估计方法。
This paper gives an estimate of correlation function for bivariate extremes Copula model using kernel regression method. A N-W kernel regression estimator is constructed and we prove that the estimator is asymptotically unbiased. Based on selection of the optimal bandwidth, we compare the N-W kernel regression estimation and OLS estimation by numerical simulation. The result shows that the N-W kernel regression estimator is more stable than the OLS estimator. So, the N-W kernel regression estimation is a relatively favourable non-parametric method.
出处
《统计学与应用》
2018年第2期234-240,共7页
Statistical and Application
基金
国家自然科学基金项目(71762008)。
