Segmenting High-dimensional Matrix-valued Time Series via Sequential Transformations
Modeling matrix-valued time series is an interesting and important research topic. In this paper, we extend the method of Chang et al. (2017) to matrix-valued time series. For any given $p\times q$ matrix-valued time series, we look for linear transformations to segment the matrix into many small sub-matrices for which each of them are uncorrelated with the others both contemporaneously and serially, thus they can be analyzed separately, which will greatly reduce the number of parameters to be estimated in terms of modeling. To overcome the identification issue, we propose a two-step and more structured procedure to segment the rows and columns separately. When $\max(p,q)$ is large in relation to the sample size $n$, we assume the transformation matrices are sparse and use threshold estimators for the (auto)covariance matrices. We also propose a block-wisely thresholding method to separate the columns (or rows) of the transformed matrix-valued data. The asymptotic properties are established for both fixed and diverging $\max(p,q)$. Unlike principal component analysis (PCA) for independent data, we cannot guarantee that the required linear transformation exists. When it does not, the proposed method provides an approximate segmentation, which may be useful for forecasting. The proposed method is illustrated with both simulated and real data examples. We also propose a sequential transformation algorithm to segment higher-order tensor-valued time series.