4
$\begingroup$

I am looking for an introductory book that helps building some skills in working with multivariate distributions.

For example, I want to be able to work with multivariate normal easily, something between an introductory stats course that talks about distributions and a serious robotics course that talks about Kalman filtering.

Improve this question
$\endgroup$
8
  • 2
    $\begingroup$ Is the question aimed at multivariate distributions or multivariate statistics? I.e. do you need more on probability or on data analysis? $\endgroup$
    – Momo
    Apr 12, 2012 at 18:50
  • $\begingroup$ What sort of things are you looking to learn? "Work with MVN" is pretty vague, and the space you describe is capacious. $\endgroup$
    – dimitriy
    Apr 12, 2012 at 19:01
  • 9
    $\begingroup$ Mark, if you are looking for continued help on CV, consider accepting some of the answers that people have provided you in the past by clicking the check mark next to one for any given question. People will be more likely to help if you aren't perceived as ungrateful. $\endgroup$
    – gung - Reinstate Monica
    Apr 12, 2012 at 19:29
  • $\begingroup$ Shouldn't this be Community Wiki? $\endgroup$
    – Carlos Accioly
    Apr 13, 2012 at 3:47
  • $\begingroup$ Johnson and Wichern is also a good reference. $\endgroup$
    – user10525
    Apr 13, 2012 at 10:42

7 Answers 7

Reset to default
7
$\begingroup$

I compiled a list of multivariate books when I was preparing multivariate statistics classes (an elective for Ph.D. students, and an intro for senior undergrads).

Improve this answer
$\endgroup$
2
  • $\begingroup$ Do you have one with lots of examples that you like? In the style of Gelman and Hill, but with a bit more rigor. $\endgroup$
    – dimitriy
    Apr 12, 2012 at 21:16
  • $\begingroup$ The best examples are in Rencher, although it is a somewhat lower level, more applied book. For you personally, Izenman will likely be the best fit. $\endgroup$
    – StasK
    Apr 13, 2012 at 13:19
5
$\begingroup$

I recommend Pattern Classification by Duda, Hart, and Stork. It's good because it is front-loaded with a lot of classical results on multivariate Gaussians, and discusses things like Fisher linear discriminant analysis, maximum likelihood, etc. The other sections of the book allow you to develop an understanding of how this connects with machine learning, which is important.

A lot of statistical texts do not make clear the overlap between classical statistics (e.g. likelihood models, logistic fitting, regression) and more modern machine learning (support vector machines, decision trees). While Duda et al. is not a great encyclopedic source for all things statistics, I think it's a good first book. It's readable and bridges gaps, which is usually the early priority before needing to chase down lots of specifics to become an expert.

Another book that might be helpful in a similar way is: Data Analysis Using Regression and Multilevel/Hierarchical Models by Gelman and Hill.

Improve this answer
$\endgroup$
1
  • $\begingroup$ Whatever its merits, Gelman and Hill is hardly a book on multivariate distributions. $\endgroup$
    – Nick Cox
    Aug 29, 2020 at 10:06
2
$\begingroup$

I can warmly recommend Michael Greeenacre's books. Biplots in Practice provides a very easy to understand overview of all kinds of biplots (concentrating on multivariate). This book is available as a collection of pdfs for free. Correspondence analysis in Practice gives a good introduction to methods such as PCA, CA and CCA. Both books contain practical explanations and codes for doing it in R.

Improve this answer
$\endgroup$
2
$\begingroup$

If your concern is 'working with the multivariate normal', I'd suggest Casella's Statistical Inference as a start.

Two classic applied first-year PhD econometrics texts are Greene's Econometric Analysis as well as Wooldridge's Econometric Analysis of Cross Section and Panel Data. These provide an overview of the theory; Wooldridge's undergraduate text also includes many worked through examples with code available online. I used Lattin in my first regression class after introductory statistics; it is filled with examples, though the code and output uses SAS.

Improve this answer
$\endgroup$
1
  • $\begingroup$ Wooldridge's texts are rightly esteemed, but they aren't focused on multivariate distributions. $\endgroup$
    – Nick Cox
    Aug 29, 2020 at 10:08
1
$\begingroup$

For the Bayesian side of things there is a really outstanding book called Bayesian Core (amazon page) with an emphasis on practice.

It begins with an introduction of grad level basics like normal distributions and quickly moves to more advanced topics like linear model, regression, mixture models etc with a lot of programming listings in R or Matlab that help grab the essential feeling of the course.

Improve this answer
$\endgroup$
1
$\begingroup$

For econometrics I used "A Guide to Modern Econometrics" by Marno Verbeek which is fairly technical but without examples.

For more applied work I sometimes use "Multivariate Data Analysis" by Hair et al. which is very practical with thorough practical examples and outputs (it lacks code for a statistical package though) including multivariate regressions, logit models, factor/cluster analysis etcetera. This one may be a little too applied for you I think and I think it's certainly an undergraduate book.

Improve this answer
$\endgroup$
0
$\begingroup$

If you have a market research background and use SPSS/SAS, you can try market research by Naresh Malhotra. It has example of code too.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.