本文在α-混合严平稳过程的假设下,研究了条件概率密度核估计的偏和均方误差.在此基础上给出了核估计的渐近最优带宽,并以S&P500指数为例展示了本文的结果.
The bias and mean square error for a kernel estimator of the conditional probability density function are studied under the assumption of α-mixing strictly stationary process. An asymptotically optimal bandwidth is given, and the results are illustrated with S&P 500 index.