4 ms·

AcceptEdge Launches College Recommendation Engine To The Public

18 points by sgrove 17y ago

physcab 17y ago
The problem that I have with this site is that it places emphasis on the numbers. My GPA, SAT, and GRE were all crap compared to my peers. But my college experience wasn't dependent on my numbers. I even wonder how indicative these scores are of future success. I agree with the author that a good college recommendation engine should show schools that you're likely to be most happy at. I realize from a data perspective that that is much harder to compute, but I feel it would be more worthwhile. Plus it seems like they are cutting off a huge portion of the market by focusing on just the numbers. What percentage of people would feel immediately discouraged by going to the site if they knew their average GPA was less than the site average? Or even more horrifying, what percentage of people will be left with the impression that applying to college would be futile in the first place?
mrshoe 17y ago
It seems (unsurprisingly) that I wouldn't have a good shot at getting into the same undergraduate program, just 10 years later. Competition seemed pretty darn cutthroat at the time; it must be a nightmare now.
skolor 17y ago
Now that, I find interesting. I graduated as top of my class, with a 4.8, got a 1490 on my SAT and a 34 on my ACT, and I only got a 310 for academics. What exactly does it take to hit the top end for that thing?
- frisco 17y ago
  we've seen that weighted gpas are usually taken apart by the colleges and don't matter as much -- it's highly dependent on unweighted gpa, though. if 10 years ago a school had 100 students applying with 4.0s, now there's over 1,000 (and we're seeing this in our dataset; it lines up with what we've been hearing from the colleges).
  - sachinag 17y ago
    Where do you get your data from? Why would the colleges want to talk to you? I'm really curious; I actually really love the idea of the service - anything that does something to reduce the uncertainty around college admissions is all right by me. (I'm assuming you're at AcceptedEdge; you really should put something in your profile. It's really quite helpful.)
    - frisco 17y ago
      Can't say too much right now, but we solve as many, if not more, problems for colleges as we do for students. If you're applying now -- please tell us what would make the site more helpful and useful! (My email's in my HN profile now.) The great thing about being an early adopter is we listen carefully and there's a good chance your suggestion will be coded up!
  - skolor 17y ago
    Ok, that's good to know. I do have one other question: You supply a "Accuracy" score, describing how accurate your predictions are. However, you supply probabilistic predictions, how do you get an "Accuracy" rating from a probability like that?
    - frisco 17y ago
      You're right that it's a probabilistic prediction, so accuracy is a little subtle. If we give 10 people chances of 70% and 7 get in and 3 don't, internally we count that we were 'correct' for all of them. For right now, the website is showing simple thresholding above/below 50% to simplify the whole thing for end-users. We're working on ways to keep complexity down while exposing the nuances of the scoring. Also, it looks like the server that computes Edge Score segfaulted this morning corrupting some of the data in memcached: I restarted it and batch re-generated scores for everyone in the system, and a lot of people that had been getting 310 / 387 scores now were showing higher scores, so that might apply to you. Just what you need when launching...
      - 17y ago
jsm386 17y ago
Setting aside the issue of relying on a recommendation engine with questionable data sources (your Facebook profile, and wherever they scraped their college data from) to influence one of the most important decisions of your life, I wonder how accurate their projections are for where you could get in, based on your grades/scores. "Your Personal Index is based on your family background, location, interests, extracurriculars, essays, social network, and other related factors. " I'd really be interested in how they weight family background. This makes a huge difference at some schools, a big difference at others, and so on and so on. Have they ranked colleges in this manner? Are they taking into account the fact that many schools are facing massive budget holes and are thus more likely to accept students who can pay their way? Source: http://thechoice.blogs.nytimes.com/2009/06/09/reed/ http://thechoice.blogs.nytimes.com/2009/06/09/reed/

18 points by sgrove 17y ago

The problem that I have with this site is that it places emphasis on the numbers. My GPA, SAT, and GRE were all crap compared to my peers. But my college experience wasn't dependent on my numbers. I even wonder how indicative these scores are of future success. I agree with the author that a good college recommendation engine should show schools that you're likely to be most happy at. I realize from a data perspective that that is much harder to compute, but I feel it would be more worthwhile. Plus it seems like they are cutting off a huge portion of the market by focusing on just the numbers. What percentage of people would feel immediately discouraged by going to the site if they knew their average GPA was less than the site average? Or even more horrifying, what percentage of people will be left with the impression that applying to college would be futile in the first place?

mrshoe 17y ago

It seems (unsurprisingly) that I wouldn't have a good shot at getting into the same undergraduate program, just 10 years later. Competition seemed pretty darn cutthroat at the time; it must be a nightmare now.

skolor 17y ago

Now that, I find interesting. I graduated as top of my class, with a 4.8, got a 1490 on my SAT and a 34 on my ACT, and I only got a 310 for academics. What exactly does it take to hit the top end for that thing?

frisco 17y ago
we've seen that weighted gpas are usually taken apart by the colleges and don't matter as much -- it's highly dependent on unweighted gpa, though. if 10 years ago a school had 100 students applying with 4.0s, now there's over 1,000 (and we're seeing this in our dataset; it lines up with what we've been hearing from the colleges).
- sachinag 17y ago
  Where do you get your data from? Why would the colleges want to talk to you? I'm really curious; I actually really love the idea of the service - anything that does something to reduce the uncertainty around college admissions is all right by me. (I'm assuming you're at AcceptedEdge; you really should put something in your profile. It's really quite helpful.)
  - frisco 17y ago
    Can't say too much right now, but we solve as many, if not more, problems for colleges as we do for students. If you're applying now -- please tell us what would make the site more helpful and useful! (My email's in my HN profile now.) The great thing about being an early adopter is we listen carefully and there's a good chance your suggestion will be coded up!
- skolor 17y ago
  Ok, that's good to know. I do have one other question: You supply a "Accuracy" score, describing how accurate your predictions are. However, you supply probabilistic predictions, how do you get an "Accuracy" rating from a probability like that?
  - frisco 17y ago
    You're right that it's a probabilistic prediction, so accuracy is a little subtle. If we give 10 people chances of 70% and 7 get in and 3 don't, internally we count that we were 'correct' for all of them. For right now, the website is showing simple thresholding above/below 50% to simplify the whole thing for end-users. We're working on ways to keep complexity down while exposing the nuances of the scoring. Also, it looks like the server that computes Edge Score segfaulted this morning corrupting some of the data in memcached: I restarted it and batch re-generated scores for everyone in the system, and a lot of people that had been getting 310 / 387 scores now were showing higher scores, so that might apply to you. Just what you need when launching...
    - 17y ago

jsm386 17y ago

Setting aside the issue of relying on a recommendation engine with questionable data sources (your Facebook profile, and wherever they scraped their college data from) to influence one of the most important decisions of your life, I wonder how accurate their projections are for where you could get in, based on your grades/scores. "Your Personal Index is based on your family background, location, interests, extracurriculars, essays, social network, and other related factors. " I'd really be interested in how they weight family background. This makes a huge difference at some schools, a big difference at others, and so on and so on. Have they ranked colleges in this manner? Are they taking into account the fact that many schools are facing massive budget holes and are thus more likely to accept students who can pay their way? Source: http://thechoice.blogs.nytimes.com/2009/06/09/reed/ http://thechoice.blogs.nytimes.com/2009/06/09/reed/