Profile Evaluation: Will GRE Math Subject Test help?

[daniel1]

Hello. Hope everyone's coping during this difficult covid19 period.

I come from a non-traditional background and would like to pursue Statistics, perhaps via a rigorous M.S. Statistics program as a first step. Would apply to a PhD program if my profile were competitive and if I had a fairly solid idea of the areas that interest me.

The following are questions I have on my mind. Would be grateful if you could share your thoughts/experiences regarding these:

- Will a good score on the GRE math subject test help?

- Will publishing at a machine learning venue help?

- Am I targeting the right universities?

Undergrad Institution: A non-US, top 25 in THES & QS rankings

Major: Computer Science

Minor: Mathematical Finance

GPA: 3.80/4.00

Student type: International

Relevant Courses:
Probability (year 1 course), Statistics (year 1 course), Multivariate Calculus, Linear Algebra, Discrete math, Data Structures and Algorithms, Quantitative Finance.
(Grades are A's and the occasional B.)
(Have not taken real analysis)

General GRE: Verbal 163 93%; Quant 169 95%; Ana 5 92%

GRE Mathematics Subject Test: should i take this?

Experience: Currently a machine learning research engineer with 3 years of experience. Trying hard but still struggling to publish at reputable venues.

Aspiration: Enter a PhD program, or seek a job whose functions are related to statistics preferably in tech (a familiar industry) as a Plan B out of academia.

Universities in my watchlist and thoughts about them (open to hearing your opinion):
California:
    -UC Berkeley (Did not apply. My profile isn't competitive for PhD. M.A. Stats program seems more for preparing students for industry.)
    -UCLA (waiting for reply)
    -UC Davis (No thesis option for M.S.)
    -USC (admitted for M.S.)

Canada:
    University of Toronto (Should I apply?)
    University of Waterloo (Should I apply?)

New York:
    Columbia (M.A. Stats. Should I apply?)
    NYU (M. Applied Stats. Don't think I should apply.)
    Cornell (No terminal masters)

 

Stay safe during this period and remember that you aren't alone in this difficult time. Thanks for reading!

[Stat Assistant Professor]

Scoring well on the math subject GRE is sometimes helpful for someone with a non-traditional background like yourself. But it wouldn't be *nearly* as helpful as getting a Masters (or just taking advanced math classes at a local university as a non-degree seeking student) or some research experience. The majority of international applicants have high scores on the subject test, and yet, a lot of them are shut out of top PhD programs in Stats. And I believe one poster (also international) on this forum was also a non-traditional applicant who got research experience while working on a Masters at Harvard, and he was admitted to Statistics PhD programs at Harvard, Duke, and Berkeley -- even without submitting the subject test score. So the subject test is by no means sufficient NOR necessary to get into a reputable PhD program in Stats. 

I don't see that you have taken real analysis or any upper division proof-based math classes besides discrete math, and your math background will be considerably lighter than the most competitive international applicants. So I would recommend that you work on strengthening your math background and seeing if you can get some research experience rather than preparing for the math subject GRE. A Masters would help you do both of those. (Note: by "research experience," I don't mean that you need to be publishing first-author papers in stat journals -- the PhD program should train you to do that. But you could volunteer some time to collaborate on a project with an established professor, so you can get a good letter of recommendation speaking to your "research potential").

[daniel1]

On 5/12/2020 at 10:55 PM, Stat Postdoc Soon Faculty said:

Scoring well on the math subject GRE is sometimes helpful for someone with a non-traditional background like yourself. But it wouldn't be *nearly* as helpful as getting a Masters (or just taking advanced math classes at a local university as a non-degree seeking student) or some research experience. The majority of international applicants have high scores on the subject test, and yet, a lot of them are shut out of top PhD programs in Stats. And I believe one poster (also international) on this forum was also a non-traditional applicant who got research experience while working on a Masters at Harvard, and he was admitted to Statistics PhD programs at Harvard, Duke, and Berkeley -- even without submitting the subject test score. So the subject test is by no means sufficient NOR necessary to get into a reputable PhD program in Stats. 

I don't see that you have taken real analysis or any upper division proof-based math classes besides discrete math, and your math background will be considerably lighter than the most competitive international applicants. So I would recommend that you work on strengthening your math background and seeing if you can get some research experience rather than preparing for the math subject GRE. A Masters would help you do both of those. (Note: by "research experience," I don't mean that you need to be publishing first-author papers in stat journals -- the PhD program should train you to do that. But you could volunteer some time to collaborate on a project with an established professor, so you can get a good letter of recommendation speaking to your "research potential").

Appreciate you taking the time to write this . Reading this, I suppose I should aim for a M.S. in Statistics rather than a PhD at the moment.

I don't think I would be able to get hold of a professor in statistics given that I didn't major/minor in stats.

Do you think that publishing in a related field like machine learning/AI be (more/less/equally helpful/ uncertain) compared to taking math classes as a non-degree student w.r.t. a MS/PhD application?
 

[StatsG0d]

1 hour ago, daniel1 said:

Appreciate you taking the time to write this . Reading this, I suppose I should aim for a M.S. in Statistics rather than a PhD at the moment.

I don't think I would be able to get hold of a professor in statistics given that I didn't major/minor in stats.

Do you think that publishing in a related field like machine learning/AI be (more/less/equally helpful/ uncertain) compared to taking math classes as a non-degree student w.r.t. a MS/PhD application?
 

It will help your application for sure, but not as much as taking the classes. Analysis is a prerequisite for most PhD programs. If it's not officially a prerequisite, it's highly recommended. Given that you are an international student, you will face stiff competition from other international students who have taken Analysis. You don't want to preclude yourself from consideration by not having the minimum background.

If you're interested in machine learning and AI and you don't want to take the math courses, perhaps you should consider a CS program.

[Stat Assistant Professor]

2 hours ago, daniel1 said:

Appreciate you taking the time to write this . Reading this, I suppose I should aim for a M.S. in Statistics rather than a PhD at the moment.

I don't think I would be able to get hold of a professor in statistics given that I didn't major/minor in stats.

Do you think that publishing in a related field like machine learning/AI be (more/less/equally helpful/ uncertain) compared to taking math classes as a non-degree student w.r.t. a MS/PhD application?
 

I think research experience does help the application a bit insofar as you can possibly get a good recommendation letter out of it. One of the posters on this forum is an international student from Australia who was admitted to Harvard, Duke, and University of Washington Statistics, and they had experience doing applied statistics in public health/epidemiology research (however, they *also* had a ton of math from a top university in Australia). You probably wouldn't do methodological/theoretical (bio)statistics research with a Stat/Biostat professor unless you are a PhD student in stat or biostat. But as a Masters student, you could definitely do some applied/clinical research and get a good recommendation letter out of that.

That said, having more math is much more important. So if you had to choose coursework vs. research, I would prioritize taking more math classes and getting good grades in them.

Edited May 15, 2020 by Stat Postdoc Soon Faculty

[movingtostats]

On 5/12/2020 at 8:25 PM, Stat Postdoc Soon Faculty said:

And I believe one poster (also international) on this forum was also a non-traditional applicant who got research experience while working on a Masters at Harvard, and he was admitted to Statistics PhD programs at Harvard, Duke, and Berkeley -- even without submitting the subject test score. So the subject test is by no means sufficient NOR necessary to get into a reputable PhD program in Stats.

Can you please tell me what masters at Harvard did this student pursue, was it biostats? Because last I checked Harvard does not have a MS statistics program.

 

[Stat Assistant Professor]

11 minutes ago, movingtostats said:

Can you please tell me what masters at Harvard did this student pursue, was it biostats? Because last I checked Harvard does not have a MS statistics program.

 

See the above profile. In Biostatistics, I am not aware of any programs that 'strongly recommend' the Math Subject GRE, let alone require it. It's just not that useful for Statistics, plus so many international applicants (particularly from China) are able to score very well on it, so it rarely "makes or breaks" an application (except at Stanford).

[movingtostats]

On 5/16/2020 at 6:38 AM, Stat Postdoc Soon Faculty said:

See the above profile. In Biostatistics, I am not aware of any programs that 'strongly recommend' the Math Subject GRE, let alone require it. It's just not that useful for Statistics, plus so many international applicants (particularly from China) are able to score very well on it, so it rarely "makes or breaks" an application (except at Stanford).

I see, thanks.

[daniel1]

On 5/15/2020 at 9:14 PM, StatsG0d said:

It will help your application for sure, but not as much as taking the classes. Analysis is a prerequisite for most PhD programs. If it's not officially a prerequisite, it's highly recommended. Given that you are an international student, you will face stiff competition from other international students who have taken Analysis. You don't want to preclude yourself from consideration by not having the minimum background.

If you're interested in machine learning and AI and you don't want to take the math courses, perhaps you should consider a CS program.

 

On 5/15/2020 at 9:39 PM, Stat Postdoc Soon Faculty said:

I think research experience does help the application a bit insofar as you can possibly get a good recommendation letter out of it. One of the posters on this forum is an international student from Australia who was admitted to Harvard, Duke, and University of Washington Statistics, and they had experience doing applied statistics in public health/epidemiology research (however, they *also* had a ton of math from a top university in Australia). You probably wouldn't do methodological/theoretical (bio)statistics research with a Stat/Biostat professor unless you are a PhD student in stat or biostat. But as a Masters student, you could definitely do some applied/clinical research and get a good recommendation letter out of that.

That said, having more math is much more important. So if you had to choose coursework vs. research, I would prioritize taking more math classes and getting good grades in them.

Appreciate the practical advice and sharing of experience.

I was due to read math courses at berkeley/nyu as a non-degree seeking student (start jan 2020), but had to shelve this course of action due to covid-19. (I'm an international applicant)

Would anyone have chanced upon a real analysis course that allows remote exams* and gives a grade?

*maybe in a setting similar to the GRE exams haha

Edited May 17, 2020 by daniel1
grammar & style

[StatsG0d]

14 hours ago, daniel1 said:

 

Appreciate the practical advice and sharing of experience.

I was due to read math courses at berkeley/nyu as a non-degree seeking student (start jan 2020), but had to shelve this course of action due to covid-19. (I'm an international applicant)

Would anyone have chanced upon a real analysis course that allows remote exams* and gives a grade?

*maybe in a setting similar to the GRE exams haha

The only one I am aware of is University of Illinois

[Stats2021]

In addition to @StatsG0d's suggestion, I understand that UC Berkeley offers Summer Sessions over the summer, and real analysis might be in the course list

 

 

Edited May 21, 2020 by Stats2021