Principal Components Analysis: A Tool for Simplifying Multivariate Data
Document ID: 2006-312 (previously 2006/006)
Published on: 1st February 2006
Author: Rutang Thanawalla
Principal Components Analysis (PCA) effectively shrinks the size of a multiviarate data analysis problem by identifying common factors or components in the data. The method is especially relevant as a first step in Barrie & Hibbert's equity model calibrations. This document goes over some of the theory (e.g. why eigenvalues and eigenvectors appear in PCA) and provides a worked example.
