A First Course in Statistical Programming

bryangoodrich

Probably A Mammal
#1
I often try to get a book from my library to help me pass the downtime at work. Lately this has been an R book (e.g., Data Manipulation, ggplot2, lattice, and others). I want to keep up on the more programming perspective of R, since I am feeling more and more confident with my ability to use R for other means (save for graphics, I just haven't had enough experience). With that said, I've made a list, and one of them is difficult for me to obtain (my local library cannot get it from any in the interlibrary loan network).

The book I'm interested in is Bruan and Murdoch's A First Course in Statistical Programming. It only has 6 reviews and 4/5 stars. The book is less than 200 pages, but covers a lot of topics. What caught my eye was that it covers linear programming--something I want to learn a little more about. It also covers some linear algebra that we skipped over in my numerical analysis class. In general, I think this could be a good book, depending on how its examples, code, and explanation are.

My (girl)friend will be able to get it for me from her school, so I'll check it out later next month. I'm curious to know if anyone has used this book and what they think about it. Anyone interested in working through the book with me?
 

bryangoodrich

Probably A Mammal
#2
For those interested, another book to look into will be Data Mining with R (2010). This is a topic I'll pick up later in my career. I just have too much on my plate for awhile now. On that note, anyone interested in that topic should also know and check out The Elements of Statistical Learning. While there are some negative reviews, the book does seem pretty good, and I've seen it referenced a bit in the literature. It's too advanced for me now, which is why I'm going to wait before I pick it up! I'll probably approach them both at the same time.
 

Dason

Ambassador to the humans
#3
Part of the problem with The Elements of Statistical Learning is that it assumes you know quite a bit. It's not a very good book to learn from - but it is a decent summary of what's out there.
 

bryangoodrich

Probably A Mammal
#4
Part of the problem with The Elements of Statistical Learning is that it assumes you know quite a bit. It's not a very good book to learn from - but it is a decent summary of what's out there.
Yeah, that is why I'm waiting to learn from it once I've learned a bit more. I don't quite have the requisite background to tackle the book, and right now it would just take too long to get up to speed. It is definitely a topic I want to get into during or after my grad school.
 

spunky

Can't make spagetti
#5
thanks! this is something i'd like to get more into as well... if you were to choose one that's the more ... uhm... "friendlier" towards introductory students in advanced statistical programming, which one would you choose?
 

bryangoodrich

Probably A Mammal
#6
thanks! this is something i'd like to get more into as well... if you were to choose one that's the more ... uhm... "friendlier" towards introductory students in advanced statistical programming, which one would you choose?
Well, advanced topics just wouldn't be accessible to introductory students! Advanced topics would cover numerical methods and the theory, which requires a background in real analysis and other mathematics (and statistics). I think it can be tackled from two perspectives: numerical methods and statistical methods. The former covers topics to be found in a numerical analysis class, such as linear solvers, integration and derivatives, interpolation, simulations, etc. The latter would cover topics like cross-validation, bootstrapping, etc. This isn't to say the two are disjoint. For instance, you use numerical methods to solve non-linear regression functions based on certain starting values, or both would deal with monte carlo simulations. The book I'm interested in looking at here seems pretty good in tackling the former from the perspective of using R. I could have pretty much used this book to do all my assignments from my numerical analysis class since it covers just about everything we covered, and I used R for all my algorithms (for an advanced treatment, I should have done more complicated tasks using C and interfacing that through R). I've found another good book that covers both topics pretty well: Introduction to Scientific Programming and Simulation Using R. I think when I come into some money, I'll add both of these books to my library.
 

bryangoodrich

Probably A Mammal
#7
I could add, two classes by a really "R guy" at the University of California, Davis (where I plan to attend) cover these: STA 141 Statistical Computing and STA 242 Introduction to Statistical Programming. A lot of the stuff he covers in this class is very useful, and often neglected by statisticians today. I'm especially going to spend time this next semester going over these assignments, especially the ones involving grabbing information from the web, using API, and producing output that can be displayed on Google Earth (which ties into the stuff I want to learn about using R with Esri's ArcGIS by interfacing with Python). One practical area I know I can apply it is that the National Center for Health Statistics apparently has an API to interface with their public mortality data. Depending on how that API works, I may be able to write code to develop queries that can automate my data collection and processing. It is a lot easier to do that than manually set up a form through their WONDER site and then export the output to a csv file and then import that into R, maybe after some cleaning (ugh!). These are the sort of things statisticians in the field today need to know how to do.
 

spunky

Can't make spagetti
#8
well, numerical methods is not something i'm unfamiliar with and monte carlo simulations are kind of the workhorse of my trade... the issue i have is that all of these were things that i was kind of taught bits and pieces as i went through my degree, so i've had to do a lot of self-teaching in graduate school now( which is not entierly a bad thing but i've just realised i learn better when there's some sort of syllabus involved. i get distracted too easily, lol). i guess what i should've said is that i need a good overall introductory book to patch the holes of knowledge i've had to kind of avoid or hop around due to lack of time/skill to teach myself how to do things, lol..