论文部分内容阅读
向量自回归模型(VAR)广泛应用在对时间相依的多元时间序列建模中,但在高维数据建模中,自回归的系数膨胀可能导致噪音估计、不稳定的预测、解释上的困难等问题。在实际应用中,序列的真实模型往往具有稀疏性,因此运用稀疏VAR模型对高维时间序列进行建模,不仅可以解决高维数据带来的上述困难,也有利于寻找高维数据内在的真实模型。本文以10家公司的股票收益率为研究对象,采用3种不同的稀疏估计方法,不但分析了股票收益率之间的动态关系,而且通过实证分析展示了稀疏估计的优势。
Vector autoregressive models (VARs) are widely used in time-dependent multivariate time series modeling, but in high-dimensional data modeling, autoregressive coefficient expansion may lead to noise estimation, unstable prediction, interpretation difficulties, etc. problem. In practice, real models of sequences are often sparse. Therefore, using sparse VAR models to model high-dimensional time series can not only solve the above difficulties caused by high-dimensional data, but also help to find the intrinsic truth of high-dimensional data model. In this paper, the stock returns of 10 companies are taken as research objects. Three different methods of sparse estimation are used to analyze not only the dynamic relationship between stock returns, but also the advantages of sparse estimation through empirical analysis.