Dimension reduction for stationary multivariate time series data
Fayed Awdah M Alshammri, Jiazhu Pan · Strathprints: The University of Strathclyde institutional repository (University of Strathclyde) · 2017
Chang et al. (2016) extended PCA by finding a linear transformation of the original variables such that the transformed series is segmented into uncorrelated subseries with lower dimensions. This method is called TS-PCA. In our current research, we will extend TS-PCA by reducing the dimension of the transformed subseries further by applying GDPCA by Pena and Yohai (2016) to the results from TS-PCA, and possibly reach a further dimension reduction. Hence, the proposed method is a combination of TS-PCA and GDPCA.