What range of schools should my profile be applying to?

[teafortwo]

I'm mostly aiming for applied math + CS programs, but statistics is definitely something I'm looking at since probability is something I'm into. Problem is I know almost nothing about statistics programs, so I'd appreciate advice from someone who does know stats

Type of Student: Domestic Asian Male

Undergrad School: Big State School

Undergrad Major: Math + CS/Econ minors

Undergrad GPA: 3.5 cumulative, 3.9 major + minors

GRE: Quant 168 - 94%, Verbal 166 - 97%, Writing 5.0 - 92%; Math Subject 650 - 48% (owwww - this is what I get for walking into the GRE without prep)

Courses: 9 grad courses: 6 in math, 2 in econ, 1 in CS. Relevant covered topics: harmonic analysis, PDEs, functional analysis, probability theory, stochastic calculus, econometrics. Also some undergrad courses of note: math stats, algorithms & complexity.

Research Experience: 3 papers. 1 is in quant finance and mostly pedagogical, so not important. 1 is in stochastic processes. 1 is in theoretical computer science. All three have been submitted to journals and are still in review, and I won't have any publications in time. This worries me a lot.

Recommendation Letters: 2 from research professors, 1 from a professor who taught two of my grad classes.

I've been told to apply for Yale and Columbia Statistics so far, but I don't even know whether they're above or below or at my level. I'd love some feedback from other people who might know the stats field well, or at least better than I do. Where SHOULD I be applying? Is my app strong? Is it weak? WTF am I doing?

Edited December 7, 2018 by teafortwo

[Stat Assistant Professor]

I think you need to have a clearer idea of what you're looking to get out of in a doctoral program, as admissions for PhD programs (and the PhD programs themselves) in each of those fields is quite different. What research do you see yourself doing? Theoretical computer science, stochastic processes, or statistical machine learning? I would recommend first deciding what you would like to focus on (CS vs. applied math vs. statistics) and then targeting mainly those programs.

1) For Computer Science PhD programs, you'll likely need to have a Skype interview with a PI (only a few math and stat PhD programs conduct interviews). It is okay if you don't have publications (though that may have changed now, with many of the top candidates having at least one conference paper)... but your application needs to be quite "research-dense" to have a shot (a slightly lower GPA seems to be a lot more forgivable in CS if you have strong research experience to make up for it). You need to demonstrate a clear research focus in your application, and so the statement of purpose/"research statement" matter a great deal. See Philip Guo's advice here about how he reviews PhD applications in computer science: http://www.pgbovine.net/PhD-application-tips.htm

2) For Applied Math PhD programs, you will need to score reasonably well on the Math Subject GRE to have a chance. There is much greater emphasis on grades in advanced math classes, and the strongest applicants will have already taken graduate-level courses in mathematics (which you have done and seem to have done well in). The statement of purpose isn't as important, and it is acceptable not to have a well-defined research focus. As it stands, however, your Subject test score is too low for most programs and you'll probably need to retake it if you go the Applied Math route. I would recommend spending a great deal of time preparing for it (most students forget a lot of the stuff they learned freshman year). Prior research experience isn't considered as essential, but letters of recommendation describing research "potential" are crucial.

3) For Statistics PhD programs, you don't need the Subject GRE for most places. The emphasis for admissions is also on grades and mathematical preparation. Statement of purpose is not a big deal (unless it's absolutely terrible), so adcoms won't expect you to have a very concrete idea of your research interests. But strong letters of recommendation describing research "potential" are crucial. Your overall GPA is a bit lower, but your math GPA is quite a bit better, and you've also taken a lot of grad level math classes, so that is a plus.

Your chances also heavily depend on how prestigious your "big state school" is. If you could give an idea of how well-regarded it is (e.g. what range of USNWR rankings), that will give us a better idea what range of schools you should target.

Edited December 7, 2018 by Stat PhD Now Postdoc

[teafortwo]

1 hour ago, Stat PhD Now Postdoc said:

I think you need to have a clearer idea of what you're looking to get out of in a doctoral program, as admissions for PhD programs (and the PhD programs themselves) in each of those fields is quite different. What research do you see yourself doing? Theoretical computer science, stochastic processes, or statistical machine learning? I would recommend first deciding what you would like to focus on (CS vs. applied math vs. statistics) and then targeting mainly those programs.

My priorities are, depending on whether the research is something the university focuses on, the following: stochastics > theoretical computer science > economic computation > scientific computation > machine learning & AI.

I'm getting away with having different interests by submitting different statements to different schools. I have a clear idea of what I want to do in each of these fields, down to the specific research groups.

Quote

1) For Computer Science PhD programs, you'll likely need to have a Skype interview with a PI (only a few math and stat PhD programs conduct interviews). It is okay if you don't have publications (though that may have changed now, with many of the top candidates having at least one conference paper)... but your application needs to be quite "research-dense" to have a shot (a slightly lower GPA seems to be a lot more forgivable in CS if you have strong research experience to make up for it). You need to demonstrate a clear research focus in your application, and so the statement of purpose/"research statement" matter a great deal. See Philip Guo's advice here about how he reviews PhD applications in computer science: http://www.pgbovine.net/PhD-application-tips.htm

So I've written up a very research-dense SOP already. Several, depending on the school. My professors are also certain our papers will be published, so I assume the subject material is nontrivial and worth discussing in my SOP. I also have research statements tailored for each program. Providing the schools I'm applying to aren't sharing my apps with each other, I should be fine. I think.

I'm a bit concerned about your comment on top candidates having a paper. What schools are they going for? What schools am I competitive in?

Quote

2) For Applied Math PhD programs, you will need to score reasonably well on the Math Subject GRE to have a chance. There is much greater emphasis on grades in advanced math classes, and the strongest applicants will have already taken graduate-level courses in mathematics (which you have done and seem to have done well in). The statement of purpose isn't as important, and it is acceptable not to have a well-defined research focus. As it stands, however, your Subject test score is too low for most programs and you'll probably need to retake it if you go the Applied Math route. I would recommend spending a great deal of time preparing for it (most students forget a lot of the stuff they learned freshman year). Prior research experience isn't considered as essential, but letters of recommendation describing research "potential" are crucial.

I have to hope my professors wrote me good letters, since two were involved in my research. The third can only attest that I passed tough grad courses, which is not good, but there's no alternative here.

Aware, sadly, that my subject GRE is trash, but I'm not waiting a year. I wasn't interested in studying either for the general or the subject, and I walked right into both blind. Anyways, I'm not aiming for the top applied math schools. So what's your thoughts on the schools I can apply to? I'm interested in seeing how your assessment matches with my professors', who I'm suspecting are overly optimistic.

Quote

3) For Statistics PhD programs, you don't need the Subject GRE for most places. The emphasis for admissions is also on grades and mathematical preparation. Statement of purpose is not a big deal (unless it's absolutely terrible), so adcoms won't expect you to have a very concrete idea of your research interests. But strong letters of recommendation describing research "potential" are crucial. Your overall GPA is a bit lower, but your math GPA is quite a bit better, and you've also taken a lot of grad level math classes, so that is a plus.

Yep. I failed a few times in totally unnecessary courses my freshman year that have no impact on anything really besides my GPA. My major and minor and general education scores are all 3.9-4.0.

Quote

Your chances also heavily depend on how prestigious your "big state school" is. If you could give an idea of how well-regarded it is (e.g. what range of USNWR rankings), that will give us a better idea what range of schools you should target.

Rank ~50.

Thanks for responding!

Edited December 7, 2018 by teafortwo

[Stat Assistant Professor]

For Computer Science, there is a bigger emphasis on "fit," i.e. whether you can fit well into a research lab. This is because much of your PhD support will be from a professor's funding (you may be supported as a TA for a few semesters, but most of the time, you'll be supported as RA). So in this case, it would be best to send follow-up emails to professors expressing your interest in their lab (whereas these types of e-mails to individual profs are typically ignored in Math and Statistics). For this reason, it is harder to assess your chances in CS than in math or statistics, because there is so much consideration given to individual research fit. 

Re: papers. It seems to be more common nowadays that top applicants have a conference publication (not necessarily as first author, but sometimes second or third author), but it may not be strictly necessary except at the top CS programs. Having research experience and three papers submitted is a very positive point for your application. 

For Applied Math and Statistics, your funding will usually be by the department, so research "fit" is not given as much attention. Admissions is more based on "hard" numbers, like GPA, math grades, and test scores. I have found that the Results page on thegradcafe.com, past years' admissions results posted on thegradcafe's "Mathematics and Statistics" sub-forum, and past years' admissions results posted on mathematicsgre.com give a pretty good picture of your chances. Strong recommendation letters also matter. If you are concerned about how much the subject GRE score affects your chances, perhaps you can talk with your math professors who might be more familiar with the admissions process. They can tell you whether or not your academic performance in graduate level math courses and your research experience will be enough to mitigate the low score.

Edited December 8, 2018 by Stat PhD Now Postdoc