Hadlye Wickham has done some amazing work in R (ggplot2, plyr and dplyr). Despite R being a bit weird and annoying at times, his packages continue to make it the most compelling option for data analysis (though Julia is catching up!).
I was just thinking today that I have very mixed feelings about Hadley, despite not knowing him -- on the positive side, his works (in my mind) transform R from a horrific language to a tolerable one, which is an astounding feat. On the negative side, if he hadn't made such momentous contributions, maybe people would have abandoned the bloody thing by now and we'd be using Julia instead.
Also my feeling. You've been a great inspiration about how to write great and useful code, but I still feel like the many great features of R (as you said some time ago, NA handling, data frames, etc.) are sometimes outweighed by the cons.
Quick question: what would you tell the Julia guys as a recommendation? Even though they are "competition", you probably have strong feelings about what features of R must be in any language that aims to replace.
What I most like about this text is that it is about the R language and not about the domain in which R has found the greatest application. I find R perfectly suitable for basic, pre-statistical signal processing for example. Far better than Matlab at a fundamental level. In fact, about anything that Matlab is good for R is better. Except of course for the evolution of the surrounding packages.
All the prior tutorials and courses I've seen presume and utilize a background in the domain of statistics and a require a fairly strong one at that. I strongly believe R should be sprung free of that encumbrance so as to find a wider audience and this is a really good effort in that direction.
As someone that likes lisp, I enjoyed a lot reading this. My first try when reading about functional programming was to try: Filter(Negate(is.numeric),c(1,2,"hello","bye")) just to receive something unexpected, then in freenode someone answered that c cast types to the more general type, here strings. So I have to begin reading from the beginning about types.
I think that Hadley can gives a very sharp opinion about what is needed to transform julia in a better R, if that is possible. Julia is about speed and no so much to make a big community, people use R because there are a lot of packages and is easy to install and well supported. Python with pandas show that you cat catchup is you try. So the question is if julia will receive some strong support or will be always a second rank players, time will reveal.
R is a lot like javascript - seems like voodoo until you 'get it' (get comfortable using it) ... and then it's the obvious choice for many common tasks.
None of the comments here really capture the importance of this book in my mind. Hadley-headed projects have dominated the data-science space for years - defining how open source and commercial platforms are expressing data processing + visualization.
This book gives you the tools to compose your own data tools using the building blocks Hadley uses. That is a big deal. Everybody should buy 5 copies.
They're basically identical. The website is updated a bit more frequently, but the updates are mainly minor wording fixes, and get incorporated whenever the book gets re-printed (every 6-12 months)
Personally I prefer the website because I can normally remember which chapter something is in, and then I can find-in-page to quickly jump to what I was thinking about.
Aah yes the old "write a book to help one remember how to do it" trick. My technical output spans a book and various places on the web. The number of times I go googling for a solution, to find an interesting text ... then I wonder who wrote it, look at the author and discover it was me. It makes me laugh and cry at the same time :)
I worked through parts of the book (especially the functional programming section). It is very well thought out and presented logically and clearly.
This has to be the effect of him doing live confresses and being a professor while also knowing the packages as only an author can.
It is also great how approachable Hadley is in the community. I still remember making a short tweet on dplyr and within 10 minutes he replied back to me and answered my questions.
I understood functional programming reading this website; thanks for that!
And ggplot2 aesthetics was the way to better understanding many of javascript plotting libraries available.
I personally still find data.table package more confortable to use than piped dplyr, but I believe "this sane competition", let's say, between packages is contributing enormously to R.
I've had the most amazing opportunity to meet Hadley at a Users group meeting. He is fantastic and very down to earth. Our group refers to his packages as the Hadley stack and I've heard that reference elsewhere as well. He has made a very powerful impact on how R has been evolving!
Listened to the FLOSS weekly podcast where they had him on (it's a bit older) recently during my commute. Recommended:
http://twit.tv/show/floss-weekly/306
Thanks, Hadley, for all the hard work.