I'm looking for reference to get a better understanding and overview of the methods for data modeling (mostly related to business questions like finding categories in data and setting of scoring functions for forecasting). After some internet research I couldn't single out a canonical reference. I have some very basic knowledge of statistics and I believe the topics that could be most useful for me would be SVM, Decision Trees, Clustering, Covariance Matrix, Correlations, Multivariate analysis ... just to name some keywords for the direction. Also I'd like to have some foundation to eventually dive into other statistical and data mining topics.

While the reference should introduce the concepts, I'd like it to be fast-paced (not chatty, not too many examples) and mathematically thorough while concentrating on applications. Of course there are specialized books on each topic, but I'd rather learn the basics of all of these topic quickly. So ideally one or two books for all of them together.

I cannot think of a different way to explain :)

Can you suggest good books? Or maybe just help me get some orientation in the vast field of statistics so that I know what to look for.

  • $\begingroup$ Have you looked at our books list? A combination of this tag with multivariate-analysis or machine-learning or data-mining should bring many references textbook, IMO. $\endgroup$ – chl Nov 1 '12 at 23:26
  • $\begingroup$ I looked at the questions with some tags. However, it seems most recommendations are quite specialized. I will try to judge if they are the right for me - however this is tedious. Maybe there is also an all-in-one starter for concepts for all these methods. Just for the basics and for me to know which method I should continue studying once I have a specific problem. $\endgroup$ – Gerenuk Nov 1 '12 at 23:43
  • $\begingroup$ @chl, SE still has an issue w/ searching under old tags. (See my meta-SO Q, eg: how can you find questions with SE's search features that have now defunct tags). Since 'books' was made a synonym for 'references', when you click on it you can only search 'references'. $\endgroup$ – gung Nov 2 '12 at 2:34
  • $\begingroup$ So is there a better tag for books? $\endgroup$ – Gerenuk Nov 2 '12 at 7:48

Maybe Elements of Statistical Learning by Hastie, Tibshirani, and Friedman? You can find it on Amazon, or Tibshirani has a free PDF download on his website. It covers a fairly broad range of modern statistical modelling methods with decent rigor and good examples.

  • $\begingroup$ @Jonathon Christensen - this book is wonderful, but you really need a good background in calculus and linear algebra to get the most from it $\endgroup$ – richiemorrisroe Nov 2 '12 at 6:48
  • $\begingroup$ Second that, but even though it is an excellent coverage of the field, I wouldn't really recommend it as a fast paced introductory text (nor is it business oriented). $\endgroup$ – pat Nov 2 '12 at 8:16
  • $\begingroup$ Oh OK. Well, it's free so I'll try to judge :) Hope there will be more suggestions. The "fast-paced" point ruled out many "standard thick" books I've already looked at. $\endgroup$ – Gerenuk Nov 2 '12 at 8:26

If you have some knowledge of R (or want to), this new book sounds like it might be useful to you.

R for Business Analytics, A Ohri.

It is a very quick and broad tutorial based overview of using R for many business analytics topics, with one chapter dedicated to data mining (using the R Rattle GUI). Don't expect a lot of formal math though.

*recent review

** It's often useful to learn a lot of the practical aspects of data mining and machine learning by learning alongside specific programming language examples. If you have a particular programming language you prefer, I could try to recommend alternatives.

***One other broad, but gentle, introduction to these areas (less practical examples, and not business based, but more mathematically inclined) is: Introduction to Machine Learning, Ethem Alpaydin.

edit: In response to your comment. If Python is your language of choice, look no further. Machine Learning, by Stephen Marsland is hands down the best practical book to pick up many of your requested topics, with many good code based illustrations to actually build and follow along.

On the Pandas side, you might have some interest in a current Coursera course (just 2 weeks in) that is Python oriented, with Python based applications towards Portfolio analysis, financial markets, and simulation backtesting (although it's a bit introductory, the instructor participates in a AI based Hedge Fund consulting firm). Computational Investing Part 1, Tucker Balch

  • $\begingroup$ I would prefer using Python. It seems Pandas is a useful addon. Yet, I hope some mathematical understanding together with some library functions will do the job. So it seems to me programming language is not so much of an issue. $\endgroup$ – Gerenuk Nov 2 '12 at 8:55

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.