Research on the nonparametric methods in time series analysis

本研究采用了经验似然比、分位数匹配等多种非参数方法来估计时间序列模型中的兴趣量。考虑的时间序列模型和概率分布有多元非高斯线性过程、局部平稳过程、稳定分布、椭圆分布等。实际应用，如最优投资组合估计问题，也考虑。

In this research, the various nonparametric methods, such as empirical likelihood ratio, quantile matching were used to estimate the quantity of interest in the time series models. The considered time series models and probability distributions were multivariate non-Gaussian linear process, locally stationary process, stable distribution, elliptical distribution, etc. Practical applications, such as optimal portfolio estimation problem, were also considered .