08-21-2019, 19:32

Author Topic: Going to a stat conference!  (Read 3347 times)

Offline k//eternal

  • hungry
  • Overdrive
  • ***************
  • Posts: 20856
    • altered.fm
Going to a stat conference!
« on: 08-02-2009, 01:10 »
It's pretty much the biggest annual conference that exists, so things should be fun. Heading out early tomorrow morning, and I'll be out for 3 days or so. See you all later :aw)

Offline Imasock

  • Honorary Asian
  • Necro Install
  • ***********
  • Posts: 6204
  • HardcoreJonny
    • http://imasock.pestermom.com
Going to a stat conference!
« Reply #1 on: 08-02-2009, 01:47 »
Later, enjoy the trip.

Offline icy is a fish again

  • And you’re still a polar bear in the jungle!
  • Overdrive
  • ***************
  • Posts: 21088
  • much moe funer
    • http://www.pestermom.com
Going to a stat conference!
« Reply #2 on: 08-02-2009, 12:36 »
Stat, as in...?

"I NEED 30CCs OF WELFARFIN, STAT!"

or

"Let me check some stats."?

Offline eternalomega

  • Anne of Green Gables erotica
  • Drive User
  • *****
  • Posts: 692
Going to a stat conference!
« Reply #3 on: 08-02-2009, 20:15 »
Should be fun...for you; it would probably be tantamount to the 4th circle of hell for me.

@icy- TRIVIA TIME: stat is short for statim, the latin word for immediately.

Offline Lenna

  • EX User
  • **
  • Posts: 104
Going to a stat conference!
« Reply #4 on: 08-02-2009, 22:16 »
Yea, i'm still wondering what a 'stat conference' is.

Is it a conference where they talk about... statistics?

Offline eternalomega

  • Anne of Green Gables erotica
  • Drive User
  • *****
  • Posts: 692
Going to a stat conference!
« Reply #5 on: 08-02-2009, 22:37 »
Or a conference that is over very quickly.

Offline Imasock

  • Honorary Asian
  • Necro Install
  • ***********
  • Posts: 6204
  • HardcoreJonny
    • http://imasock.pestermom.com
Going to a stat conference!
« Reply #6 on: 08-03-2009, 01:05 »
Well, considering that k// is a biostatistician, it would be a conference that relates to the field of statistics in some way.

Offline eternalomega

  • Anne of Green Gables erotica
  • Drive User
  • *****
  • Posts: 692
Going to a stat conference!
« Reply #7 on: 08-03-2009, 01:41 »
I got that, but...the jokes...they call to me.

PS that is a fancy title.

Offline Imasock

  • Honorary Asian
  • Necro Install
  • ***********
  • Posts: 6204
  • HardcoreJonny
    • http://imasock.pestermom.com
Going to a stat conference!
« Reply #8 on: 08-03-2009, 01:54 »
Not much longer than a mathematician, though much more difficult for 99% of people to say.

Offline Antronach

  • Quantity has its Quality
  • Armagus User
  • ****
  • Posts: 317
  • ==Uber DIE Mode==
Going to a stat conference!
« Reply #9 on: 08-03-2009, 02:28 »
But the bio part would make it something like "bodily mathematics", wouldn't it?
a banner

Offline Imasock

  • Honorary Asian
  • Necro Install
  • ***********
  • Posts: 6204
  • HardcoreJonny
    • http://imasock.pestermom.com
Going to a stat conference!
« Reply #10 on: 08-03-2009, 14:35 »
Kind of.  Biostats = the field of statistics using biology, and encompasses such things as biological experiments, most notably in medicinal and agricultural applications; the collection, summarization, and analysis of data from those experiments; and the interpretation of, and inference from, the results.

Offline k//eternal

  • hungry
  • Overdrive
  • ***************
  • Posts: 20856
    • altered.fm
Going to a stat conference!
« Reply #11 on: 08-04-2009, 00:00 »
Yeah, what Sock said, basically.

Fun stuff so far, and yeah, we're talking statistics here. Typing from iTouch so I'll get back with details later. Still there for a bit.

Offline Imasock

  • Honorary Asian
  • Necro Install
  • ***********
  • Posts: 6204
  • HardcoreJonny
    • http://imasock.pestermom.com
Going to a stat conference!
« Reply #12 on: 08-04-2009, 00:13 »
Cool beans, =P

Offline icy is a fish again

  • And you’re still a polar bear in the jungle!
  • Overdrive
  • ***************
  • Posts: 21088
  • much moe funer
    • http://www.pestermom.com
Going to a stat conference!
« Reply #13 on: 08-05-2009, 11:22 »
Word, any new distributions to be on the lookout for?

Btw, eternalomega, I loves your double extras. :3

Offline k//eternal

  • hungry
  • Overdrive
  • ***************
  • Posts: 20856
    • altered.fm
Going to a stat conference!
« Reply #14 on: 08-05-2009, 20:24 »
Well, new distributions, I didn't see any, those are kind of rare events but I wasn't on the lookout, either. Mostly I was looking for techniques related to work in one way or another.

There was one guy exploiting properties of graphics processing units to conduct a phylogenetic analysis in about 3 orders of magnitude less time, which was pretty neat. The last speaker we saw was also really cool, he had a method he called GPS (ostensibly it stood for generalized path seeker, but everyone knew it was intentionally a pun) to simultaneously obtain parameter estimates across an entire class of penalties for penalized linear regression. Pretty crazy, but I didn't get to see how he was going to select the final model, because we had to go bowling and stuff (birthday party, not mine though).

Anyway! Fun few days, good to be back, how is everyone doing :aw)

Offline icy is a fish again

  • And you’re still a polar bear in the jungle!
  • Overdrive
  • ***************
  • Posts: 21088
  • much moe funer
    • http://www.pestermom.com
Going to a stat conference!
« Reply #15 on: 08-05-2009, 23:27 »
Huh, do you know of something that would help you to, say, find the most interesting points by assigning it a characteristic, and allow you to take care of that also dynamically (e.g. finding trendlines above a certain range)?  That was my problem this summer, and I didn't know if there were some things in statistics that might help on storage for that purpose.

Offline k//eternal

  • hungry
  • Overdrive
  • ***************
  • Posts: 20856
    • altered.fm
Going to a stat conference!
« Reply #16 on: 08-05-2009, 23:57 »
Hn, I'm not understanding your question as much as I would like to in order to answer it. Is the characteristic (1.) something real that's already known, (2.) something real that's unknown, or (3.) something that's just made up and consequently known?

1. Well, in this case, you don't really have a problem at all, so let's move on.

2. In this case, I suppose your problem is trying to match new points whose "interesting" status is unknown based on information you also have for the known points. That's a pretty standard classification problem.

3. I'm not sure what kind of application this might have... maybe if instead of having the actual parameter values in your training set, you only have something that approximates it and therefore probabilities instead of 0/1, so you attempt to solve the problem by simulation? In that case, just generate several training sets using the known probabilities to obtain binary values, and proceed as in #2.

I'll toss you the PDF that I wrote up on classification a while back via PM if this is what you mean, otherwise, more elaboration plox 8D

Offline icy is a fish again

  • And you’re still a polar bear in the jungle!
  • Overdrive
  • ***************
  • Posts: 21088
  • much moe funer
    • http://www.pestermom.com
Going to a stat conference!
« Reply #17 on: 08-06-2009, 13:16 »
The dominated set problem, i.e. Find all points greater than or equal to the starting point s.t. the set of dominating points are all as good as other points in the set (whether le or ge) and better on at least one dimension.  Equivalent points fit into this dominating set.

This is a database problem, and I'm wondering if after you have the data in the database, you can make this characteristic so that it's more useful than just importing all of the data.  One solution sorted the data based on entropy values (ln(product(each value))) each time, and it made it behave roughly as a one dimension-less problem.  In any case, when you make the characteristic, the data is known, but the data that comes in will be an unknown.  Does that clarify the problem for you at all?  So, depending on the application, it may be #1, #2, or #3, depending on "what works". XD

I'd be happy to see the PDF.  I'll send you my paper at this point if you'd like to see it as well. :3