A0463
Title: Sparse principal component analysis for high-dimensional stationary time series
Authors: Yan Liu - Waseda University (Japan) [presenting]
Kou Fujimori - Shinshu University (Japan)
Yuichi Goto - Kyushu University (Japan)
Masanobu Taniguchi - Waseda University (Japan)
Abstract: Sparse principal component analysis (PCA) is explored for high-dimensional stationary processes. Traditional PCA is ineffective when the dimension of the time series is high. Oracle inequalities are presented for penalized PCA estimators covering a broad range of stochastic processes, including those with heavy tails. The convergence rates of these estimators are established, along with theoretical guidelines for selecting the tuning parameter. The performance of sparse PCA is illustrated through numerical simulations. Furthermore, the practical utility of sparse PCA is demonstrated using average temperature data.
