I'm a current student at NCSU working on an evidence-based advising platform (Quantios) that can predict high school students chance of getting into college (acceptEdge). I just started a survey that is populating my database and was wondering if anyone would be up for helping me out with some data points on your admissions information. (I have a few $50 amazon gift cards to give away in a drawing for filling out the survey).

Survey Location: http://survey.quantios.com/?offer=_amzn_50

Thanks!

5/18/2009 3:15:26 PM

interesting project. can you tell us more about the kind of technical analysis you are doing or is this a commercial/proprietary system?

also, normally it's bad etiquette to cross post threads, but you'll probably get more response if you post this in chit chat. in fact, if you want me to move it there, let me know.


[Edited on May 18, 2009 at 3:23 PM. Reason : though this thread title sucks for chit chat, so just remake it there if you want]

5/18/2009 3:21:09 PM

Appreciate it! We will be launching the Facebook application at the end of this month, and the beta external website a few weeks after.

5/18/2009 3:23:25 PM

Here's a screen shot of the external website as it sits on our development server.

5/18/2009 3:51:20 PM

Most universities, including NCSU, keep very detailed statistics on their applicants and can usually predict with a pretty good level of accuracy how well their students will perform including being able to predict their GPAs within a certain margin. I remember that my adviser in FYC my freshman year had various kinds of data like that in my records including a very accurate estimate of my final undergraduate GPA.

5/18/2009 3:51:20 PM

Darkone: That's great insight, I had no idea they were predicting GPA's as early as high school. We take into account literally every piece of data we can get our hands on. GPA / SAT / ACT / AP's / Honors / Legacy / Athletics etc.

5/18/2009 4:00:53 PM

qntmfred: sure I wasn't sure on the policy about cross-posting threads, if you can delete one of these threads for me that'd be great!

5/18/2009 4:31:13 PM

Looks solid. How are you guys planning on making money?

5/18/2009 5:11:11 PM

purdy

i hate tabs on the side though

5/18/2009 5:16:04 PM

muell: We allow students to fill out a objective / subjective profile, and we basically run regressions on current admissions data using clustering to figure out how far the student is from the mean. We then package up all this information in a good looking PDF file and sell it back to the student for $2.99 price range.

Bigman157: I agree I'm not a huge fan of side-tabs, it's a solid free-feedback system we found, but I'll eventually change it to something less obtrusive (and ultimately take it off once the beta is done)

5/18/2009 5:34:21 PM

That's a good idea... I see you have an "essays" tab, are you guys writing essays for people

5/18/2009 6:11:30 PM

Ya'll going to be considering race, home location, school applying to, etc... in this analysis?
Very interesting concept.

5/18/2009 6:50:58 PM

We also take into account race / home location / parents average income, here's a another screen shot of the Facebook app we have been developing alongside of the external site:

5/19/2009 1:50:58 PM

How do you deal with people entering in erroneous data??

5/19/2009 5:18:17 PM

Quote :

"How do you deal with people entering in erroneous data??"

why are you asking this? aren't you involved with the project...judging from your thread in the classifieds (now moved to trash bin)

5/19/2009 5:23:20 PM

yup recently introduced to the whole 'data mining' aspects, this way everyone can be familiar with how the system works

5/19/2009 5:30:15 PM

the point is, don't be disingenuous about it. asking planted questions like you're not affiliated reeks of lame. don't ruin a potentially interesting project with shady PR

5/19/2009 5:34:15 PM

I'm actually curious about the statistical tools that you'll use for this project. I imagine that the complicating factors of non-gaussian data distributions and all sorts of covariance make this sort of thing an interesting challenge.

5/19/2009 6:19:20 PM

why does everyone assume that they're even doing proper analysis of the data

just playing devil's advocate

5/19/2009 6:21:34 PM

^ Maybe I asked the question in order to probe their competence in tackling such problems?

5/19/2009 6:24:13 PM

ditto

5/19/2009 10:10:44 PM

HaLo, at this point, I could probably take any set of raw data, do some analysis on it, have the outcome be statistically valid, and, use the outcome to prove or disprove any side of any statement or argument. Sad, but, true. Hence the reason methods are so important to be scrutinized. (The same holds true for accounting ... can manipulate all sorts of things on the books and have it all be perfectly legal/valid).

I agree with darkone. The U.S. News and World report rankings are often questioned because they use a "secret" weighting algorithm of sorts. Credibility of the underlying methods is going to be key for buy-in to the website. Love the idea, but, myself would be very suspicious upfront.

This looks interesting and would be curious to see how it progresses over time. Just don't run out of cash while pursuing this.

5/20/2009 9:39:54 AM

Quote :

"would be very suspicious upfront"

i'm already suspicious due to the manner in which they asked questions of themselves making them appear they were coming from an unaffiliated party...AND the fact that the "planted" questions were asked by someone who was talking about the project in another section of this site. i didn't even have to look at the IP logs to identify them as someone involved with the project. if you're gonna be dishonest, at least put some effort into it

5/20/2009 10:24:41 AM