February 12, 2008 4:58 PM   Subscribe

I am looking for great books / guides for learning applied statistics and R (at the same time). Any hivemind suggestions?

I am a graduate student in political science. This summer I have 114 days of no work, no research, and no classes (yeah!). Along with studying for my comprehensive exams, I want to spend this time getting a good handle on applied statistics and the program "R".

Little Background:

I have weak background in mathematics but a strong background (and proficiency) in programming.

My first statistics classes were heavily "math / theory" oriented (proofs, formulas, etc...). I had trouble grasping it and really did not enjoy myself.

However, in the more advanced classes we started using R and I loved it. To me, statistics in R is just enough programming language to learn / tinker in, something I am very comfortable with. I throughly enjoy spending time hacking code and so learning statistics through R seems like an interesting idea.

This is all just a long winded way to asking: Does anyone knows of some great introductory, intermediate, or advanced books that teach applied statistics through using R?
posted by chrisalbon to Technology (13 answers total) 24 users marked this as a favorite
Data Analysis Using Regression and Multilevel/Hierarchical Models and Modern Applied Statistics With S (don't worry, R is an implementation of S) are good places to start. Neither book is really strong on teaching stats theory, so taking a look at Blumer's Principles of Statistics might be a good idea.
posted by thisjax at 5:35 PM on February 12, 2008 [1 favorite]

A fellow grad learning stats here. I've just come across Simple R—Using R for Introductory Statistics, which I'm using to cover the R side as a course text (David S. Moore, The Basic Practice of Statistics) covers the stats.

Related problem: R is ungooglable. Too many other pages about stats have that letter on them somewhere. Any tricks?
posted by eritain at 5:50 PM on February 12, 2008 [2 favorites]

I COMPLETELY feel your pain about Googling "R".

The only trick I know is to use Google Code Search. In the advance setting you can select "R". I have found it useful to search for examples of code or (my favorite) search the comments in commented code.
posted by chrisalbon at 6:17 PM on February 12, 2008

Argh...was just coming on here to recommend the same two books that thisjax did...though not in that order for learning R. Modern & Applied Statistics With S ("MASS") is awesome and, IMHO, has yet to be replaced as the all-purpose goto book for R. Gelman's book (the other one recommended by thisjax) mixes use of R and BUGS, but its beauty is in its smart and clear explanation of cool methods -- it's one of those books that will help you continue to fall in love with data analysis.

I'm also quite partial to Frank Harrell's Regression Modeling Strategies, which does what too few stats books do -- shows you how to think like a serious analyst in terms of how to organize a complete analysis and how to work with real-world (read: messy!) data.

As far as good google-fu for R goes: try r-seek (which also has a search bar extension for Firefox, as does the r-help mailing list, and the r-wiki).
posted by shelbaroo at 6:29 PM on February 12, 2008 [2 favorites]

for googling r stuff append site: r-project.org to searches about how to implement various techniques in R (packages) - for bonus points use your local cran mirror instead of the central site. for mailinglists google site:tolstoy.newcastle.edu.au (and other places where the archives live).

here's a particularly good document I came over the other day which talks about differences between R and other packages implementations.

Oh and so that you know, R is actually a variant of lisp, don't let some of the c-like syntax fool you :)
posted by singingfish at 7:16 PM on February 12, 2008 [2 favorites]


I am also interested in learning a more general programming language to help me gather data. I am been looking into Python. Any suggestions out there? Anything that works especially well with R?
posted by chrisalbon at 8:15 PM on February 12, 2008

See also the sidebar here for other books.
posted by eritain at 9:42 PM on February 12, 2008

... aaaand, once you know the statistical methods you want to use, the quick and dirty guide will get you there. Now I'm done. I think.
posted by eritain at 9:51 PM on February 12, 2008

If you want an R book I highly recommend An Introduction to S and S-Plus by Phil Spector, which works just fine for R. It will give you a brief introduction to the language and its idiosyncrasies, but not much applied statistics. It is very short though, and you might find learning stats from another book easier after reading it. Unfortunately most of the online introductions I have seen suck, as do several of the books with R in their name.

I use Python and R almost exclusively. I think they work pretty well together. I try to do my heavy lifting in Python, and then summary analysis or graphics in R. Write out data with headers using tabdelim in the textinput package (self-link; you can use the built-in csv instead if you don't mind a little more setup). Then read it into R with read.delim().
posted by grouse at 11:57 PM on February 12, 2008 [1 favorite]

I am also interested in learning a more general programming language to help me gather data.

I'd suggest that you join the polmeth list and ask them whether you should be learning python or perl or what, given your particular needs and goals.

The big reason I'd suggest this is that you want the opinions of other people who primarily do political science but need to program from time to time, not so much the opinions of people who program for a living and dream code. Different goals, different needs might mean that different packages are most useful.

If nothing else, I'd be interested in seeing people's replies.
posted by ROU_Xenophobe at 11:59 AM on February 13, 2008

Once again Xeno you offer me great advice.

posted by chrisalbon at 12:39 PM on February 13, 2008

Happy to help, even though last time I confused my Jackmans and thought that you'd be taking a methods comp from *Simon* Jackman, the thought of which makes me want to piss myself in terror.

If you're not on polmeth, it really is an excellent and normally quite friendly resource. You might also find a programming language discussion in the archives, but I don't remember one offhand.
posted by ROU_Xenophobe at 7:26 PM on February 13, 2008

Hee - I wrote the quick and dirty guide mentioned above. That server is down now, but the current version can be found here
posted by redbeard at 2:05 PM on October 29, 2008

« Older Help me make delicious turnovers   |   Life changing movies Newer »
This thread is closed to new comments.