Principal Component Analysis (PCA) is a mathematical tool that can capture the most important information (variance) in data. Recently, CMS collaboration applied the PCA technique to analyze the data in Pb+Pb collisions, and a non-zero subleading mode has been extracted and interpreted as arising from subleading eccentricity . In this talk, we study the performance of PCA method in the study of the factorization breaking effects of harmonic flow in two-particle correlation analysis . We show that the choice of weighting factor in the orthogonality criteria, and $p_T$ range of the analysis, and even the statistical uncertainty of the input data could lead to significantly different results in the extracted single-particle flow modes. Therefore, it is necessary to understand the sensitivity of PCA procedure to these issues before one could attach any physics interpretation of the subleading flow modes obtained.
 CMS Collaboration, Phys.Rev. C.96.064902
 A. Mazeliauskas and D. Teaney, Phys.Rev. C91 (2015) no.4, 044902
 Z. Liu, A. Behera, H. Song and J. Jia, in preparation.