Phd Eval Stat/Biostat

[Danieldm]

Student type: Domestic white male

Major: Mathematics BS, Computer Science BA at a top 20 statistics and top 10 biostatistics school

GRE: 169 V, 169 Q, 5.0 W, no subject tests

GPA: 3.5 Overall, 3.4 Math Major, 3.75 CS Major

I know it's not good but I do at least have an upward trend... GPA will be a 3.7+ all semesters in my Junior and Senior years, which includes a full load of summer classes in addition to normal semesters.

Research Experience:

One project that will be submitted to ICASSP 2020 and will be accepted/rejected sometime during the application process, so they may see it or may not. 9 months of work on deep learning for drug classification, state of the art result on a popular dataset in the subfield. Professor does not do machine learning, but is a moderately well known theorist. Also dept. chair.

One project that may be submitted to a conference around the time of applications but will not be accepted or rejected at that point. It's on using recurrent neural networks to forecast the cognitive abilities of Alzheimer's patients on their next clinical visit. 6 months of work. Professor is fairly famous, 10,000+ citations, well known ML researcher in our biostatistics department. Also dept. chair.

I will be first author of two on both these papers, I did most of the work and came up with the research questions. 

Courses:

I'm only gonna discuss the super relevant ones.

B+ in real analysis, A- linear algebra, A in intro graduate/advanced undergrad ML in the stats dept., A in intro graduate/advanced undergrad ML in the comp sci dept., and some grad courses. My university uses a weird scale and it converts to undergrad grades harshly... explained in my letters of rec. Pass/B in most programs/C here in intro to statistical theory, High Pass -/A- in most programs/B+ here in intro to statistical theory 2, high Pass -/A- in most programs/B+ here in numerical analysis. The statistical theory sequence is well known to be brutal and is considered the hardest pair of graduate courses in both of our statistics departments, so that'll be reflected in my letters.

Letters:

2 from the guys I'm doing the research projects with, one from the prof I took intro to statistical theory 2 and the machine learning class in the stats dept. from. All should be quite strong, both people I am doing research with said they wanted to see me end up at Carnegie Mellon or MIT.

Master applied to:

Just UNC-Chapel Hill MS in statistics. I can transfer credits and do it in a year, but this is ONLY if I don't get into any PhD programs.

PhD Programs:

Statistics:

Harvard, Carnegie Mellon, UM-Ann Arbor, UW-Seattle, University of Pennsylvania

Biostatistics:

UNC-Chapel Hill, Harvard

Is this a reasonable list given my stats? I know my GPA is low but all else seems to be good. I was aiming for the 6-12 range for programs.

Edited October 21, 2019 by Danieldm

[Stat Assistant Professor]

Unfortunately, I think your GPA is too low for the PhD programs on your list, even with the grade deflation/unusual grading scale that you mentioned. It may be somewhat forgivable if your math GPA was much better than your overall GPA or if you attended a school known for its grade deflation (e.g. UChicago or Reed College), but I have a hard time seeing you getting admitted into any of those schools. Your list is far too top-heavy and I would advise you to apply more widely.

[Danieldm]

If I don't get in, I'll just go do the masters and reapply in a year.

Also if it matters, I do have the upward trend going. GPA will be a 3.7+ for my junior and senior year, which includes a full summer load in between time.

[bayessays]

If you'd like to do a PhD, I'd recommend also applying to more reasonable programs now.  Even if you got an MS with a 4.0 and good research I still wouldn't bet on getting into the programs you listed.

That being said, the research is really a wild card.  If I'm reading between the lines correctly here, you wrote a paper with and are getting a letter from one of the top couple biostatisticians in the world.  Your grades are way better than mine and I got into some programs similar to the lower end of your list, but my results were all over the place.  I wouldn't be totally shocked if you got into the biostat programs you listed. But I also think you could very easily be rejected from a school in the 30-40 range. 

[Danieldm]

I basically got that same spiel from my profs. Actually, they encouraged me not to apply to places much worse, because they thought that the further down you go, the more heavily GPA is weighted. They both said I'd have a better chance to get into a program ranked 10/15 than a program ranked 30/40. I'm kind of a wonky applicant so I'm interested to see where I get in. I basically think this is super high variance, and the more people I talk to the more that seems to be the view.

Thanks! I'll be sure to post an update on where I end up in a few months.

Also you said you got into some programs similar to the lower end ones that I listed... may I ask where?

Edited October 21, 2019 by Danieldm

[bayessays]

UNC biostat/other top 5 biostat programs.

[Stat Assistant Professor]

I would have to agree with bayessays. I think even with a MS, your chances at the Statistics PhD programs on your list are not very good (not sure about Biostat -- maybe your chances there are better, since they do seem to highly value methodological/applied publications more than Stat). These programs are super difficult to get into -- especially UPenn Wharton which matriculates only 4-6 students every year, and only one or two of those will be domestic students. I don't see schools like Harvard or Penn taking a chance on someone with your GPA. I would apply much  more broadly than only the top-tier programs. With a strong performance in a Masters program (if you go that route), your chances are probably pretty good at schools like UFlorida, Ohio State, etc.

Edited October 21, 2019 by Stat PhD Now Postdoc

[Danieldm]

Sounds like I oughta apply to more biostats then...

Should probably knock off Penn for it. You're right that they admit basically nobody given the small size.

Edited October 21, 2019 by Danieldm

[Stat Assistant Professor]

5 minutes ago, Danieldm said:

Sounds like I oughta apply to more biostats then...

Should probably knock off Penn for it. You're right that they admit basically nobody given the small size.

If you are more interested in methodological/applications research (especially applications in public health, genomics, etc.), then I would suggest applying to more Biostatistics programs than Statistics. You can also work on applied stuff in Statistics departments, but I would say in general, Stat programs are a bit more focused on statistical theory than Biostat (outside of the tip-top Biostat programs).

[Danieldm]

I basically narrowly care about three things. 1. ML for healthcare/precision medicine, 2. Architectures for statistical learning that mimic the human brain/operate more like real intelligence and 3. robotics.

I'm also applying to 4 CS programs. May up biostatistics to 3 and add in Ann Arbor.

But yeah... theory is by and large not my cup of tea. 2. is closest, but that's kinda more of a neuroscience/coding problem than a math problem.

Edited October 21, 2019 by Danieldm

[bayessays]

Definitely agree with Postdoc in that I think Harvard/Penn/UW statistics are a waste of money, and there is probably not much of a path for you to get into places like that in the future.  But I got waitlisted by Chicago after doing well in a MS program, so I'm not going to tell you it's impossible - I don't think Chicago is nearly as difficult to get into as those places though.

I would definitely add Michigan to your biostat list and see if there are any other biostat programs with people that interest you for research interest #1, which I think you'll find mostly in biostat departments.  For interests 2 and 3, you'll pretty much be looking at the CS places obviously.  

[DanielWarlock]

I'm also applying this year. Your profile is comparable to mine, i.e. with more emphasis on research rather than grades. I know people who are on the opposite end with perfect grades in a bunch of graduate math classes (harmonic analysis, Lie algebra etc.) but no research experience or letters to testify. Their letters will be from math class instructor to further reiterate that they are "math geniuses". Yes on paper, they appear to be more "solid". But the admission needs to balance research interests of incoming cohort. And that is where they fall short. The purpose is not to fill the entire cohort with machine learning oriented people but is to ensure professors in each area will have some, even just remote matches, even more so they didn't admit any in previous years. For example, if there are +2 sampling, statistical computing, Markov models faculties, then they will very likely admit at least one person with experience and interests in that area. That way, your disadvantage compared to someone from pure math on baseline measures who think he is going to do machine learning or "big data" or theoretic probability in general will not be as bad. Also, competition between profile like yours and mine will be almost non-existent because your area of interests are completely orthogonal to mine. So I think the best strategy for profile like ours is specificity: talk about your area in statement competently, say by proposing research ideas in relation to your target faculties' sub-sub-field. 

But at the end of the day, they want to make sure you have as much competence with mathematics as possible. Personally, I think competence with math is crucial to publication in statistics. Some profs believe in that strongly but some do not actually care as much. The consensus at Harvard where I did my masters seems to be that probability/statistics is not math and a "probabilistic" thinking is more important. Entering PhD cohort at Harvard is not exclusively math geniuses--some do struggle with 1st year probability and inference classes believe it or not. I believe there are many applicants who are extremely competent with math but didn't get in because match of interests and other factors.

That said, at top places like Stanford, you will still fall short because regardless of your area, some applicants' track record is good enough to apply to post-docs in his/her specialized area. (take a look at this guy: https://faculty.wharton.upenn.edu/wp-content/uploads/2016/11/CV_Bhattacharya.pdf?_ga=2.148566066.443551298.1571676683-1391157221.1571513002). Solving 4 world-class combinatorics problems and published in top math journals 2nd year in college.. I mean, I wouldn't be too surprised if he got a tenor-tracked position at lower-tier department fresh out of college. How do you compete with this stuff? You can have 100% in GRE math and take all math classes there are and still pale in comparison to such people. For example, I like to say my interests match Diaconis as I also worked on Markov chains but so do many other people and there is no way I can get in no matter what. And mind you there is going to be at most 2 such pure math geniuses to be admitted--they will absolutely admit people with less impressive baseline measures (if they have to) in other areas such as biostats, high-dim inference because there is only one Diaconis.  My strategy is therefore to apply to 2-3 less elite places such as UF where there are several faculties that match your interest, and leave the rest to fate (or God if you are religious).

Edited October 21, 2019 by DanielWarlock

[omicrontrabb]

I graduated with a BS in statistics and a math minor from a top 15 stats school. I had a 4.0 GPA and similar GRE scores as you and was rejected from all of those stats programs you listed (but accepted into some of the biostats ones.)
 

Your research experience is definitely exceptional, so you may have better luck than me, but you should probably also apply to some lower ranked schools.