Archived Recommended books on statistics for research (statistics)

submitted ago by JeffreyARobinson

Search Voat (via searchvoat.co)

Submission Info

Posted by: JeffreyARobinson

Posting time: 3.7 years ago on

Last edit time: never edited.

Archived on: 2/12/2017 1:51:00 AM

Traffic stats

Views: **272**

Score

SCP: **1**

**1** upvotes, **0** downvotes (**100%** upvoted it)

- Greasetrap [O]

Archived Recommended books on statistics for research (statistics)

submitted ago by JeffreyARobinson

Cookies help us deliver our services. By using our services, you agree to our use of cookies.

## Sort: Top

[–] TelescopiumHerscheli 0 points 0 points 0 points (+0|-0) ago (edited ago)

"Basic Statistics" by David Blackwell. Old, and you can only get it second-hand, but the best. Written by one of the world's greatest doctoral supervisors - the man knew how to teach, and how to encourage others to great success.

[–] admiral_pinkbeard 0 points 0 points 0 points (+0|-0) ago

What is your background?

[–] JeffreyARobinson [S] 0 points 0 points 0 points (+0|-0) ago

B.S. in Computer Science, another in Mathematics. M.S in Computer Science, going for PhD. in Computer Science.

[–] admiral_pinkbeard 0 points 1 points 1 points (+1|-0) ago

Well, as far as what you should learn goes, I think that some fundamentals in statistics and propability theory would be beneficial. Here is a text for free for a probability theory course. I only took a glance at it, but someone with a bachelor's math background should be able to understand it without much trouble. This text is a bit more formal than what you will generally use, but it is extremely useful if you want to understand assumptions being made when performing statistical tests.

After you are familiar with those basics, you should probably learn linear regression and generalized linear regression. They are two fairly straightforward tasks that can be used to help answer (or approximate) common problems.

The field of statistical analysis diverges and overlaps quite a bit from there, so it's hard to say what else would be useful without knowing your area of research. There are a lot of machine learning algorithms that are useful in computer science, although it's easier to understand the "common sense" part of the algorithms rather than the arcane statistical processes behind them, especially when the data is high-dimensional.

As a side note, from what I've seen, a lot of people in the statistics field "understand" the algorithms they are using, but they generally tend to use pre-written (usually optimized) libraries to do so, but they are not as good at writing the algorithms themselves. You have an advantage with a math/CS background, so when you are familiar with a wider variety of algorithms, you should be able to optimize them more easily.