Transition to Stats PhD

[theKiev]

I will be finishing my MS in Operations Research from Georgia Tech next fall, and would like to apply to some Statistics PhDs. I'm primarily interested in: Random Matrix Theory, Stochastic Analysis (possibly Rough Path Theory), Machine Learning Theory, and High-Dimensional Statistics (so still kind of broad).

Background Info:

Undergrad Institution: Big State School (90 - 100 range in USNews)
Major(s): Mathematics

Minor(s): Computer Science, Economics

GPA: 3.72 (both cumulative and major, give or take)
Type of Student: Domestic/White/Male

Grad Institution: GaTech
Concentration: Operations Research (MS)
GPA (so far): 4.0

GRE General Test:

Q: 169
V: 163
W: 5.0

GRE Subject Test in Mathematics: 880

Programs Applying: Statistics, several math/applied math/OR programs probably.

Research Experience:

• Undergraduate: 
  • ~7 Months in the Mathematics Department developing software tools to help several professors visualize some of the concepts they were researching (somewhat like these).
  • Summer internship in Natural Language Processing.
• Post-undergraduate: Spent about 1 1/2 years at a top 5 business school as a Research Assistant. 1 project resulted in a paper in finance (not a co-author, but my name in a footnote if that means anything).
• Graduate: 
  • Internship at a well-known tech company this summer.
  • Will be working in a lab group starting in August for a few months till I finish.

Awards/Honors/Recognitions: Phi Beta Kappa, some small recognitions/scholarships here and there.
Letters of Recommendation: 2 from research advisors, 2 from professors who taught graduate classes.

Math/Statistics Grades:  

• Undergraduate: Probability (A), Math Stats (B-), Analysis I (A-), Analysis II (A), Econometrics (A-), Operations Research (A-), Machine Learning (A), Intro to Proofs (B+), Calc III (A-), Linear Algebra (A).
• Graduate PhD taken (all As): Linear Optimization, Nonlinear Optimization, Real Analysis I, Probability I, Stochastic Processes I & II (more applied, but there are proofs), Theoretical Statistics, Probabilistic Graphical Models, High-Dimensional Statistics.
• Graduate PhD to take: Probability II, Simulation Theory, Algorithms (CS theory course)

Schools: NCSU, Washington, Chicago, Duke, Berkeley, Columbia, UCLA, and Yale (small, I know, but a professor of particular interest there).

Misc: Given my mediocre undergraduate grades, I'm not sure if there are schools I should add/delete from the above list. Ideally, I’d like to keep it in the 5-8 range. Some of my advisors are certain that some schools will place quite a bit more weight on my undergraduate record versus grad classes/research/gre, but they are a bit vague on which programs those are in terms of Statistics PhDs (obviously I'd like to minimize my applications to such schools).
 

[omicrontrabb]

Your undergraduate grades are fine. A lot of applicants will have a B or two on their transcripts. But regardless, your math GRE score is super impressive. I doubt anyone will question your ability to do advanced math if you scored in the 90th percentile for the subject test. My understanding is that score would look good at many math PhD programs.

Based on your profile, you will almost certainly get into NCSU. As for the other schools, you should be competitive, but the admissions are just so selective. I would recommend applying to some lower ranked schools as well.

It sounds like you have very specific research interests, so I would just pick programs at different tiers that do that research. It sounds like you’re mainly interested in probably theory, so I would suggest, for example, UNC instead of NCSU. Cornell also has people working in many of those areas and is less selective  than Berkeley/Chicago.

[bayessays]

Agreed, besides HD stats, your research interests are very niche (and more common at only top departments), so you'll have to choose wisely. On your list, I think you'd get into NCSU and UCLA and be in contention at the others.  I agree that I'm not sure NCSU (and UCLA really) is the best place for you given your probability interests.  UNC and Rutgers are some other reasonable targets with some research overlap.  A little lower in the rankings, Michigan State has a department that really fits your research interests but you won't need to go down that far. 

[Stat Assistant Professor]

Agreed that UNC-STOR seems like a better fit than NCSU and UCLA based on your research interests. I think your chances at University of Washington are also decent, and that program would be a very good fit for you as well. There are other schools that have some faculty working in your stated areas of interest (e.g. Tienfeng Jiang at University of Minnesota and Garvesh Raskutti at University of Wisconsin), and I think you would be a competitive candidate at UMN and UWisc.

[theKiev]

21 hours ago, omicrontrabb said:

Your undergraduate grades are fine. A lot of applicants will have a B or two on their transcripts. But regardless, your math GRE score is super impressive. I doubt anyone will question your ability to do advanced math if you scored in the 90th percentile for the subject test. My understanding is that score would look good at many math PhD programs.

Based on your profile, you will almost certainly get into NCSU. As for the other schools, you should be competitive, but the admissions are just so selective. I would recommend applying to some lower ranked schools as well.

It sounds like you have very specific research interests, so I would just pick programs at different tiers that do that research. It sounds like you’re mainly interested in probably theory, so I would suggest, for example, UNC instead of NCSU. Cornell also has people working in many of those areas and is less selective  than Berkeley/Chicago.

 

6 hours ago, bayessays said:

Agreed, besides HD stats, your research interests are very niche (and more common at only top departments), so you'll have to choose wisely. On your list, I think you'd get into NCSU and UCLA and be in contention at the others.  I agree that I'm not sure NCSU (and UCLA really) is the best place for you given your probability interests.  UNC and Rutgers are some other reasonable targets with some research overlap.  A little lower in the rankings, Michigan State has a department that really fits your research interests but you won't need to go down that far. 

 

3 hours ago, Stat PhD Now Postdoc said:

Agreed that UNC-STOR seems like a better fit than NCSU and UCLA based on your research interests. I think your chances at University of Washington are also decent, and that program would be a very good fit for you as well. There are other schools that have some faculty working in your stated areas of interest (e.g. Tienfeng Jiang at University of Minnesota and Garvesh Raskutti at University of Wisconsin), and I think you would be a competitive candidate at UMN and UWisc.

Thank you everyone for the sound advice, and specific programs/individuals to consider. Very much appreciated!

Edited May 8, 2019 by theKiev