High-dimensional low-rank tensor autoregressive time series modeling

Modern technological advances have enabled an unprecedented amount of structured data with complex temporal dependence, urging the need for new methods to efficiently model and forecast high-dimensional tensor-valued time series. This paper provides a new modeling framework to accomplish this task via autoregression (AR). By considering a low-rank Tucker decomposition for the transition tensor, the proposed tensor AR can flexibly capture the underlying low-dimensional tensor dynamics, providing both substantial dimension reduction and meaningful multidimensional dynamic factor interpretations. For this model, we first study several nuclearnorm-regularized estimation methods and derive their non-asymptotic properties under the approximate low-rank setting. In particular, by leveraging the special balanced structure of the transition tensor, a novel convex regularization approach based on the sum of nuclear norms of square matricizations is proposed to efficiently encourage low-rankness of the coefficient tensor. To further improve the estimation efficiency under exact low-rankness, a non-convex estimator is proposed with a gradient descent algorithm, and its computational and statistical convergence guarantees are established. Simulation studies and an empirical analysis of tensor-valued time series data from multi-category import-export networks demonstrate the advantages of the proposed approach.

Automated summary

This paper proposes a new modeling framework for modeling and forecasting high-dimensional tensor-valued time series using the autoregression method. By considering a low-rank Tucker decomposition, this method can flexibly capture the underlying low-dimensional tensor dynamics, achieving dimension reduction and multidimensional dynamic factor interpretations. The paper also studies different estimation methods and their non-asymptotic properties under different low-rank settings.