Learning Bayesian Statistics

bryangoodrich

Probably A Mammal
#1
Since I've joined Talk Stats, I have noticed Bayesian statistics come up a few times. I don't expect to take a class on it until I get to grad school in at least a couple of years, but I wouldn't mind getting some background on the subject in the mean time.

Does anyone have any book recommendations on Bayesian statistics? I'm talking about a good one you would buy because you want to keep it in your library. Also, any preference for ones dealing with Bayesian statistics using R? In addition, if you know of any scholarly articles that might give a good review (e.g., this one) or talk about important details for a student of the subject, I would love to see those recommendations, too.
 
#2
I havent started learning Bayesian yet either but think i will soon. Looking forward to seeing what recommendations are posted. @bryangoodrich you are going to kill it in grad school if you are already this interested in stats. props.
 

Dason

Ambassador to the humans
#3
I could tell you what books (one by Hoff and one by Alberts) we used in my Bayesian methods course but I don't think it would help too much. They were alright but to tell you the truth I didn't really use them so I can't offer too much commentary on them. I mainly learned from the lecture notes of my "Advanced Statistical Methods" course and from my "Bayesian Methods" course.
 

bryangoodrich

Probably A Mammal
#4
I saw the latter one on Amazon when I was looking up R books this summer to check out. From what I've seen, the R books don't usually give a complete treatment of the theory. So, like your course, I think I would want to combine the R book with a good theory book. I'll see if I can get Hoff's from my library now ... and it looks like I can. I might put them off until Winter recess, though. I have an upcoming project that's going to require some hardcore ArcGIS training for the rest of my summer, on top of the math review for the next semester and possibly taking the math GRE if I feel bold.
 

bryangoodrich

Probably A Mammal
#5
I havent started learning Bayesian yet either but think i will soon. Looking forward to seeing what recommendations are posted. @bryangoodrich you are going to kill it in grad school if you are already this interested in stats. props.
The hardest thing for me right now is making sure I've got the mathematical sophistication. I've been working with epidemiologists for the past 3 years, and I have really developed my R skills. I haven't put my focus on statistics because I always wanted to do something else. I've stepped back from economics and don't see philosophy as a viable alternative. I figure I can get all my enjoyment out of statistics while making loads of $$$ to take care of my family when that time comes. My focus on stats has really been a recent turn of events! I just happen to have a very, very .... very long education lol Let's just say, I graduated the class of 2000 and have been in college since ;)
 

Dason

Ambassador to the humans
#6
I saw the latter one on Amazon when I was looking up R books this summer to check out. From what I've seen, the R books don't usually give a complete treatment of the theory.
See if you have a decent background in frequentist theory then the Bayesian theory isn't that bad. It's more the general philosophy that is harder to accept the first time learning Bayesian methods. You're working with a different concept of probability than you do in the typical frequentist setting. And that can take some getting used to.
 
#7
i hear you on the long educational route. im very much still at it with absolutely no end in sight. sighhhhhhhhh. i suppose i need to come to the realization that there simply is no end.
 

Dragan

Super Moderator
#8
The hardest thing for me right now is making sure I've got the mathematical sophistication.
I think that this is the most important point for you to consider.

You should have a basic 3 course sequence in Calculus, Linear algebra, Diff (Eq), Intro to Math-Stat, Adv. Calc., Numerical Methods, Modern Algebra


At the Grad level Courses in: Math-Stat, Adv. Linear Algebra, Probability, and Stochastic Processes.
 

bryangoodrich

Probably A Mammal
#9
Yeah, I have a bachelors in mathematics. I know what to expect. It's just there is so much time in between, so I get a little rusty. I did most of my calculus in my head back in the day. Then after I got my AS in mathematics, I didn't touch an integral for many years until mathematical statistics! The masters program at UCD only requires the undergraduate mathematical statistics courses, so I'll retake it there. It will still be more advanced and in-depth than what I learned at CSUS. This coming year (at CSUS), I plan on taking the one semester of grad stats they'll offer (Spring) and a stochastic processes course (Spring, also). I'm going to try and tutor stats courses and try to squeeze applied linear algebra in, but I may end up with just abstract linear algebra due to scheduling and work. I've got a background in numerical methods and will also take linear programming next semester.
 
#10
My favorite book is
O'Hagan, A. and Forster, J. (2003) Kendall's Advanced Theory of Statistics, Volume 2B: Bayesian Inference. Arnold, New York. ISBN 0-340-52922-9
it's concise, reads very well (although sometimes you have to think hard. Start at the first page, and everything falls into place. This is the book that is more a reading than a reference book, but maybe others will disagree. In any case it's great, I always have it around.
 
#11
I was "intoduced" to this while working on a project this past winter. It was pretty embarassing to not know more about Bayesian methods so it's on my agenda as a topic to try and get up to speed on.

My favorite book is
O'Hagan, A. and Forster, J. (2003) Kendall's Advanced Theory of Statistics, Volume 2B: Bayesian Inference. Arnold, New York. ISBN 0-340-52922-9
it's concise, reads very well (although sometimes you have to think hard. Start at the first page, and everything falls into place. This is the book that is more a reading than a reference book, but maybe others will disagree. In any case it's great, I always have it around.
I looked this up on Amazon... not cheap!!! Only one copy is less than $100!!!
 

Dason

Ambassador to the humans
#12
Textbooks never tend to be cheap...

On the bright side for me - we have that in our department's reading room.
 

bryangoodrich

Probably A Mammal
#13
That's why I just "rent" books from the library (or inter-library loan program). I should take my friend's advice, though, and fancy myself some thrifty websites that might have e-copies ;) I only buy books now if (1) I can get them cheap enough, and (2) They are worth having in my library for future reference and use. Now, if anyone wants to get me a copy of Casella and Berger Statistical Inference for my birthday, it would be much appreciated! haha
 
#14
It's been a few years since I was in school... and the last year and a half of grad school I didn't buy a single book. I'm pretty anti-academia so it's been a while since I've even had to think about how much text books cost but some copies of this were over $200, I never had to contend with that!
 

Dason

Ambassador to the humans
#15
Casella and Berger is good. I think I was able to get a copy for $40 which for a textbook isn't too bad...

I don't know of any pdf version floating around but I do have a set of solutions (which admittedly doesn't provide the best or whole solutions for many problems).
 

bryangoodrich

Probably A Mammal
#16
Casella and Berger is good. I think I was able to get a copy for $40 which for a textbook isn't too bad...

I don't know of any pdf version floating around but I do have a set of solutions (which admittedly doesn't provide the best or whole solutions for many problems).
The solutions in pdf format?? I'll take those! lol $40 is pretty **** good, because the latest edition is over $100 list (Amazon) price. I'm going to add it to my save list and keep an eye on the used copies. I'll only pay for it if it goes under $50. It is definitely one I want in my library.
 

Dason

Ambassador to the humans
#17
I'd sell you mine but I don't think I've ever parted ways with a math/stats book. I LOVE THEM TOO MUCH.
 

spunky

Can't make spagetti
#19
Now, if anyone wants to get me a copy of Casella and Berger Statistical Inference for my birthday, it would be much appreciated! haha
oh wow.. did we all learn with casella and berger!? i still keep mine in a cherished place right next to my videogame walkthroughs... i didn't know it was so popular, but i always considered to be a little jewel... what about buying it used? or roaming outside prof's offices? sometimes they just dump books in these boxes with a huge signs saying FREE BOOKS and all of us the scavengers of the department (aka grad students) swarm around and leave the box empty clean... :)
 

Dason

Ambassador to the humans
#20
I'm sure they are out there but I don't know a masters program that doesn't teach out of Casella and Berger.