You are here

Understanding Singular Vectors

A certain weighted average of the rows (and columns) of a non-negative matrix yields a surprisingly simple, heuristical approximation to its singular vectors. There are correspondingly good approximations to the singular values. Such rules of thumb provide an intuitive interpretation of the singular vectors that helps explain why the SVD is so effective in analyzing large data sets.

Identifier: 
http://www.jstor.org/stable/10.4169/college.math.j.44.3.220
Subject: 
Rating: 
Average: 3 (2 votes)
Creator(s): 
David James and Cynthia Botteron
Cataloger: 
Daniel Drucker
Publisher: 
College Math. Journal, Vol. 44, No. 3 (May 2013), 220–226
Rights: 
David James and Cynthia Botteron

Comments

ddrucker@wayne.edu's picture

This is a follow-up to the other College Math Journal article coauthored by David James. It contains an interesting case study involving Major League Baseball.

meade's picture

This is more of an application of statistics that happens to involve SVD, than it is about the SVD. There are some good insights, but I doubt this would be appropriate for general use in most linear algebra courses. Could be a good topic for a student project - but more likely for a statistics course than a linear algebra course. NOTE: Accessible through JSTOR with institutional membership, or for purchase.