I have some free time this summer and would like to read something to prepare for the "Advanced Data Analysis" course I will be taking. This second level course, according to the course catalog, covers "alternatives to ordinary least squares, influence and diagnostic considerations, robustness, special statistical computation methods." A graduate-level regression analysis course (which I have already taken) is prerequisite. There will be no required text for the class.

I would also consider viewing free online lectures or taking free online courses if there are any recommendations in that area. Please keep in mind that this is a second-level graduate course, which means many online resources I've found (especially free courses and video lectures) are too basic to be of use.


One of the best is Cosma Shalizi's Advanced Data Analysis from an Elementary Point of View available for free download here ... http://www.stat.cmu.edu/~cshalizi/ADAfaEPoV/ It's eventually going to see the light of day as a published book. He pretty much covers the waterfront in terms of modern advanced analytic techniques that extend the classic multivariate framework.

A great reference for the classic techniques is Dillon and Goldstein's Multivariate Analysis: Methods and Applications which was published in 1984 but has the advantages of being highly readable as well as applied. It's a great book.

Of course, Hastie and Tibshirani are largely responsible for propagating the panoply of post-modern, advanced analytic techniques via a whole series of roving workshops back in the 90s. Their book is available for free download here ... https://web.stanford.edu/~hastie/local.ftp/Springer/ISLR_print1.pdf

There are many more references worth obtaining:

Frank Harrell's Regression Modeling Strategies

Max Kuhn's Applied Predictive Modeling Of course, Kuhn developed the caret package in R.

Gelman and Hill's Data Analysis Using Regression and Multilevel/Hierarchical Models is a great introduction to Bayesian modeling techniques and it's very hands-on.

Leskovec and Rajaraman's Mining of Massive Datasets introduces some terrific techniques for machine learning analysis of truly massive data.

There's a ton more stuff worth digging into. This should keep you busy this summer though.

  • $\begingroup$ Thank you! the Shalizi looks great. Until now, for my preparatory reading this summer, I have not been considering any books in this area written as long ago as the 90's -- or, generally, prior to the advent of R. Does that seem misguided? $\endgroup$ – Ceph May 18 '16 at 18:59
  • 1
    $\begingroup$ Not misguided but all of the post-modern advanced techniques have their roots in the classic multivariate framework. Many of those tools still see wide use. Too many data scientists today are myopic about how, e.g., SVM relates to linear discriminant analysis -- a classic tool. Moreover, papers are still being written and published that leverage these approaches -- it's not all about advanced models for statistical learning. That's just what grad programs are teaching since they want to appear up-to-date and informed about the latest big thing. $\endgroup$ – Mike Hunter May 18 '16 at 19:13
  • $\begingroup$ Along the same lines as @DJohnson, I'd say it's never a bad idea to go back to some of the fundamentals. I'm not an economist, but Greene's Econometric Analysis and Simon and Blume's Mathematics for Economists are never more than an arm's length away from my desk. $\endgroup$ – 5ayat May 19 '16 at 8:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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