2

This is kind of a follow-up question from this post: Gradient descent vs lm() function in R?

Is there any literature available for the QR decomposition concept involved in the lm() function in R?

Community
  • 1
user3532629
  • 21
  • 2
  • 3
    It is a pretty common decomposition in linear algebra [QR Decomposition](http://en.wikipedia.org/wiki/QR_decomposition) – bdeonovic Apr 21 '14 at 01:04

1 Answers1

5

As Benjamin has commented (+1) the QR decomposition is extensively used and studied regarding its use for solving least squares problems. Book-wise I have used the following books in the past:

  1. Matrix Computations by Golub and van Loan (the reference book on the subject)
  2. Numerical Methods for Least Squares Problems by Bjõrck (quite computational and to the point but strictly speaking not statistically minded)
  3. Matrix Algebra From a Statistician's Perspective by Harville (great book that probably covers all essential parts of dealing with matrices a statistician should know)

I think these books will probably cover most aspects one might care for regarding the usage QR decompositions for linear models. Please note that all three books mentioned do assume some basic knowledge of linear algebra (nothing hard-core); if you do not have that I recommend going through/looking up Introduction to Linear Algebra by Strang. It is quite manageable and gets you up to speed with what you will need for more advanced applications without being overly mathematical about it.

Ah, and this thread: Reference book for linear algebra applied to statistics? might also come in handy when looking for references.

Community
  • 1
usεr11852
  • 33,608
  • 2
  • 75
  • 117
  • Do you also have some good reference for generalised linear models that you suggest? – Steve Sep 23 '14 at 10:46
  • 1
    @Steve: In the past I have used as references the books: [An Introduction to Generalized Linear Models](http://www.crcpress.com/product/isbn/9781584889502) by Dobson and Barnett and [Generalized Linear Models](http://www.crcpress.com/product/isbn/9780412317606) (uber-classic) by McCullagh and Nelder. I also really liked the exposition of GLMMs in [Generalized Linear Mixed Models: Modern Concepts, Methods and Applications](http://www.crcpress.com/product/isbn/9781439815120) by Stroup. I have nothing "heavy" on Lin. Algebra though. – usεr11852 Sep 23 '14 at 19:23