Nonlinear PCA Based on Data Transformation

• Activity Number: 518
• Type: Contributed
• Date/Time: Wednesday, August 4, 2010 : 10:30 AM to 12:20 PM
• Sponsor: ENAR
• Author(s): Mehdi Maadoliat*+
• Companies: Texas A&M University
• Keywords: PCA ; Data transformation ; Dimension reduction

Abstract

Storage and analysis of high-dimensional dataset is a challenge. Dimension reduction methods are usually used to reduce the data while keeping most of the information in the data. PCA is one of the commonly used dimension reduction techniques. However, PCA does not work well when there are outliers or the data distribution is skewed. Therefore nonlinear dimension reduction methods have been developed in the literature. In this talk, we present a new nonlinear PCA method that first transforms the data to reduce the skewness of the data distribution and then perform the standard PCA. Our method is cast into a profile likelihood framework for efficient computation. Our method is illustrated using several simulated data sets and a real data set.