Re: Computer for Statistical Analysis
- From: "Jay Cole" <JTCole@xxxxxxxxx>
- Date: 18 Oct 2006 05:20:58 -0700
Your friends were talking about things that would not matter to you too
much. Transfering data between system RAM and the processor cache is
something we don't really care about too much unless you are doing some
intensive iterative processing and the like. Then, you are looking at
some very expensive platforms.
For setting up a computer to crawl the web and collect data, any of the
newer computers will do. I guess it depends on how many different
programs you will have running at once, and if you are writing your
software to take advantage of parallel processing, which I don't think
you would be. So, in that case a pentium 4 with one gig of ram, on
board graphics, etc. should be plenty. You're talking about a 500
computer. The only thing I would make sure I'd get is a large hard
drive and maybe even two set up using raid 1 so you are less likely to
lose data on a hard drive failure. You can call this an anti-server I
guess lol.
wojciech@xxxxxxxxx wrote:
Hi everyone,
Thank you so much for your answers, I really appreciate them.
While I do appreciate and use the standard stats packages today (I use
R for most things, and SAS for courses that require it), the reason I
have to make my own statistics software at times mainly deals with two
things:
1) I do a lot of social network analysis, and make my own software for
it, since current tools don't really do what we require. You can see
more at www.egotistics.net
2) In the next 6 months, I'll be working on data mining projects and
modelling projects and will be building crawlers and bots for the
process. The computer will be running 24/7 to collect data and keep
track of changes. This is also why I can't use my laptop! :)
Before posting on this group, I consulted my computer science friends,
who began discussing the important difference between data retrieval
from a CPU's cache and RAM, which quickly went over my head, so I
posted here to see if any of you guys had any any opinions -- for
example, statistical analysis using a dual core versus single core and
whether, in the long run, it makes any difference at all.
Ideally, I was hoping for an article from scientific-computing.com or
something that might target the "stats guy with a limited budget", much
like those video game magazines have for gaming systems... But I guess
there's a difference in the target markets here. :)
Thanks,
Wojciech
kenneth_m_lin@xxxxxxxxxxxxx wrote:
What software are you using?
<wojciech@xxxxxxxxx> wrote in message
news:1161062956.121226.17360@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
Hi,
I'm currently a student who works on a great deal of data intensive
projects, both within research labs and of my own initiative. I run a
lot of analysis on my laptop and recently decided to set up a dedicated
data analysis system. Now, I have a limited budget, and was wondering
if anyone could advise me (or point me to a site or other resource)
about what I should be looking for and how I can optimize the system in
terms of computing power and budget.
So far I'm looking at getting a Dual Core system with 2 GB of RAM, and
recycling an old grapics card and hard drive. Any advice?
Thank you,
Wojciech
.
- References:
- Computer for Statistical Analysis
- From: wojciech
- Re: Computer for Statistical Analysis
- From: kenneth_m_lin
- Re: Computer for Statistical Analysis
- From: wojciech
- Computer for Statistical Analysis
- Prev by Date: Re: m00es, Illywhacker, O'Hara, and Afonso are like Internet Cafe attandants
- Next by Date: Buffer Management
- Previous by thread: Re: Computer for Statistical Analysis
- Next by thread: Re: Computer for Statistical Analysis
- Index(es):
Relevant Pages
|
