B-School Predictor

Good Stuff..Thanks.

um it doesnt take into account your major / your ug school / where you worked / what you've been doing etc, otherwise helpful i guess

Is this a joke? This has to be some of the dumbest shit ever posted here.

It's good to know that my buddy's 4.0 in communication from a community college and his 5 years of work experience at Enterprise rent-a-car beats a 3.9 in physics from a target with 3 years of MBB.

Good to know that if I work for another 10 years, I'm a shoe in for top schools.

I think Veritas or someone else has a very good one right now?

Absolutely useless until you add the shit listed above.

Version 0.30 beta released: http://mbapredictor.com/

Updates: - Added Quality of Work Experience Input - Added Undergraduate Institution Prestige Input - Enhanced penalties for very little work experience, very poor GPA, and very poor GMAT at the absolute level (i.e. not relative to a particular school) - Corrected bug that sometimes miscalculated work experience when compared to the average matriculant at a school - Adjusted the formula for final calculation of the approximate % chance at a school

Obviously, things still need to be adjusted further, particularly the quality and prestige aspect. Other input factors may be added in the future.

It didn't even ask if I was white.

You need to give some idea of what the scales mean. 0 - 5 for work experience is meaningless if a 0 is no work experience then a 1 is enterprise rent a car or something then what is a 3, or a 4, or a 5? That is a shit ton of variance.

Some advice:

I suggest you tweak the results to produce significantly lower probabilities of getting into the top schools. The difficulty of getting in scales substantially within the top 3 and top 7. Few people in this world have a >75% chance of getting into Stanford, let alone a >95%. You'll never be able to produce a useful tool If you're running the same formula to evaluate a candidate's potential for getting into each school.

With 2 years of work experience, my chance at HSW are ~40%, but with 4 years I'm >95% at Wharton...

And with 40 years of rent-a-car work experience, I have a >95% chance of getting in everywhere even with a 1.0 GPA from a community college and a 600 on GMAT.

I suggest including x^2 along with x as regressors (I assume you're doing some sort of regression analysis), or at least just use log(x).

I don't think its terrible (since everyone is shitting on your I thought I might give some positive comment so you don't deepthroat a handgun).

The thing is, particularly with BSchool, there is no formula. Thats the whole point right? Thats why people worry about work experience and volunteering and all that other shit.

I'd say keep working on it but realize that there are just some flat out limitations to what you're trying to do so don't go into thinking it'll be as accurate as a Law School predictor or something similar.

my .02 cents

unless columbia got that much harder to get in to, the avg gmat listed is wrong (739) I think it closer to 709 or 719

i think the 'bashing' is actually really good constructive criticism to tweak it better

Re: Work Experience - I believe there's a difference between how I think it's calculating work experience and how it's actually calculating work experience, because it should already be essentially capping how much benefit you get out of lots and lots of years of work experience. I'll look over it further. Once I adjust the overall formulas for the top schools so that the chances go downwards, this should hopefully alleviate some of the problems, too.

Re: Columbia GMAT - You are correct; there's a typo. It should be 712 based on the dataset (current version of USNWR) I'm using. It'll be fixed in the next update. Thanks for the catch.

Re: Predicting B-School Admission - I don't believe that there's any way under the current system of b-school admissions to produce a prediction tool that can be as accurate as Law School Predictor (I have higher hopes that Med School Predictor, after a few rounds of updates, can be pretty accurate). After all, many law school publish formulas weighting LSAT scores and GPAs.

That said, I think with fixes/updates and some time, MBA Predictor can become a useful gauge, or starting point if you will (which, to some extent, Law School Predictor is still just a starting point).

Other: I will check in again next week to catch up on glitches and suggestions (I'll be travelling starting tomorrow through early next week and also need to take care of some administrative and tax paperwork). I'll work on the Predictor during that time, but just wanted to alert people that I'll still be looking out for feedback ... I just won't be responding until next week.

Your feedback has been helpful in identfying current problems with the beta version and in providing ideas on how to improve on it.

There's no way one can predict this ... even if there was a formula (and there is not), no applicant would possibly be able to calibrate themselves on all the factors that go into admissions.

Would be interesting to backtest this on WSO folks who are willing to supply admit decisions and inputs. You probably won't get a sample size of any significance but it would still be cool to see how well the predictions correlate with outcomes this season.

what data set are you using for this? (where did you get the data?) im assuming you are doing a logistic regression for this? i would suggest (as someone mentioned before) x^2 for work experience, log(x) for GPA/GMAT (i.e. diminishing returns).

right now its a joke but would be interesting to see if you can produce somethign usable from this