A1895
Title: Estimating the number of significant components in high-dimensional principal component analysis
Authors: Zhixiang Zhang - University of Macau (China) [presenting]
Abstract: A new penalized approach for estimating the number of significant components in high-dimensional principal component analysis is proposed, using the explained variance ratio and the rigidity of the nonspiked sample eigenvalues of sample covariance matrices of variables. Compared with methods in the existing literature, consistency of the proposed estimator holds not only for independent data but also for some time series data when the dimension and the sample size both tend to infinity. Even for independent data, the estimator works under weaker conditions than existing approaches such as AIC and BIC, including allowing heterogeneity in the bulk of the population eigenvalues. Simulation studies are conducted to illustrate the performance of the proposed estimator.
