PhD Statistics Chances (Interested in Machine Learning)

[mcaleste]

Ok here are my stats, I'm currently a senior and would love to get some feedback on what schools I should be aiming for. 

 

Undergrad Institution: Large state school (not well known for math)
Major: Pure Math and Economics
GPA: 3.76 (overall), 3.90 (upper division math)

Type of Student: Domestic White Male

Courses/ Background:

Quantitative Courses:

Calc 1,2,3 (C+ in calc 3, As in others), Intro Linear Algebra (A), Intro Abstract Algebra (A), Linear Programming (A), Applied Statistics (A), Econometrics (A+), Applied Linear Algebra (A), Intro Proofs Class (A), Advanced Calculus (A), Complex Analysis (B+), Group Theory (A), Topology (A), Graduate Real Analysis (this semester), Graduate Statistical Machine Learning (this semester)

 

I'm proficient in Python and competent in Matlab

 

GRE: 164Q, 169V, 5

 

Research Experience: Have been doing machine learning research with two professors. Currently leading my own research projects. No publications, but will give presentations to the research groups 
Awards/Honors/Recognitions: Dean's List, on pace for Magna Cum Laude
Pertinent Activities or Jobs: Two Wall Street internships (non-technical), founded startup, two software engineering internships, research assistant, Tutor, leader of multiple student orgs

Letters of Recommendation:

-Two professors I'm currently doing research with

-Finance professor who heads the student org I lead, have worked with him over multiple semesters

Thinking of applying to:

CMU

Duke

Wisconsin

University of Washington

Columbia

UCLA

 

Do I have a shot at any of these schools? 

[mcaleste]

Would appreciate any advice. 

[cyberwulf]

I don't think it's implausible that you could be admitted to a couple of schools on your list (I would think chances are better at Wisconsin/Columbia/UCLA than CMU/Duke/UW), but I would recommend adding some more places ranked in the 15-25 range as safeties.

[StatsG0d]

Your profile is pretty similar to mine when I was applying (research experience, grades, etc.). Frankly, I think the list of schools you put is much too top-heavy. I applied to UCLA and Columbia of those schools you listed, and was ultimately waitlisted to UCLA and rejected from Columbia (but accepted for their non-funded MS). As cyberwulf mentioned, you'll want to add in at least a few schools in the 15-25 range. Sometimes results can surprise you (both in good and bad ways).

[mcaleste]

How about Cornell, UT Austin, Penn State and Yale? What schools did you end up getting into and would recommend for me as safeties? I'm specifically focused on machine learning research. 

[GoPackGo89]

Would you, @cyberwulf or @footballman2399 mind elaborating on why @mcaleste isn't a top applicant? He has great grades in upper level math classes and assuming he gets an A in graduate real analysis there is another. Decent GRE scores, research experience which I am assuming will lead to great letters. I have looked at past profiles on this site for a year and mcaleste looks great to me. Just wondering what I am missing

[cyberwulf]

12 minutes ago, statbiostat2017 said:

Would you, @cyberwulf or @footballman2399 mind elaborating on why @mcaleste isn't a top applicant? He has great grades in upper level math classes and assuming he gets an A in graduate real analysis there is another. Decent GRE scores, research experience which I am assuming will lead to great letters. I have looked at past profiles on this site for a year and mcaleste looks great to me. Just wondering what I am missing

Don't get me wrong, I think the overall profile is strong. However, there are a couple of lower math grades (e.g., C+ in Calc 3), and the overall GPA suggests a couple of lower grades in other courses as well. In addition to the strength of the letters, a lot of @mcaleste's success will depend on the strength and reputation of the "large state school" he attends. If it is a major state flagship (like, say, University of Virginia or similar), then this profile may be enough to get him into a top-tier program. Otherwise, probably not.

[StatsG0d]

Of those schools, I'm confident you'd have a good shot at Penn State. Yale is (very) difficult to get into. I think they prefer that applicants take (and do well in) the math GRE.

I'm not too familiar about the other programs to give you information. 

Regarding the other question, I think it depends on how forgiving committees are with the B+ in Complex Analysis. In my experience and from those I've heard about, it is extremely difficult to crack the top 7, but after that it becomes significantly easier. Letters could make a big difference though. 

[Robbentheking]

@footballman2399

does anyone in stats really care at all about complex analysis? I could imagine some of the techniques for integration might pop up in some proof somewhere, but otherwise it doesn't seem too applicable.

[Residuals]

1 hour ago, Robbentheking said:

@footballman2399

does anyone in stats really care at all about complex analysis? I could imagine some of the techniques for integration might pop up in some proof somewhere, but otherwise it doesn't seem too applicable.

Probability theory can use a lot of complex analysis topics. For example, characteristic functions.

[StatsG0d]

2 hours ago, Robbentheking said:

@footballman2399

does anyone in stats really care at all about complex analysis? I could imagine some of the techniques for integration might pop up in some proof somewhere, but otherwise it doesn't seem too applicable.

I think at the very top they do care. Depending on what you're learning, there's not too much from even Real Analysis (I) that's helpful in probability. However, it shows one's ability to do theoretical and abstract math, so regardless of whether it's "useful," in terms of a prerequisite, it's useful as a tool for admissions. 

I'm by no means saying the OP doesn't have a shot, but it's not a shoe-in either. That's why I would advise the OP to apply to a few top 7 schools as well as a few (non-Ivies) in the 8-20 range. 

[Residuals]

Real analysis will lead one from undergraduate-level coverage of probability to graduate-level coverage and formulation of probability using measure theory. Also, as I said, complex analysis is useful for a number of important probability-based approaches. So, knowing real analysis for theoretical statistics is more than a litmus test of being able to do math for admissions committees, it is necessary for having a firm grasp of a number of theoretical concepts in statistics.

Edited November 23, 2016 by arima

[mcaleste]

To provide a bit more information, I go to Arizona State University (so no public ivy). The C+ in calc 3 was first semester freshman year where I already knew most of the material from high school so I stopped studying/going to class but I didn't know the last section and didn't study enough in time. The complex analysis class was a dumb non-proof-based class, so I didn't prioritize it highly. As for total GPA, other low grades came almost exclusively from business classes and pre-requisites (marketing, management, etc.) Not sure if that helps, but thought I would mention. 

[StatsG0d]

Unfortunately, I don't see how that could help. The committees might be forgiving on Calc 3 if you say something like it was taken freshman year and it was difficult to get used to the college lifestyle of something (of this I'm not certain). But there's not going to be much of a defense for Complex Analysis. I would advise against saying you didn't study for a class on your personal statement to justify your grade. 

I'm confident you'll get in somewhere, just be rather cautious about where you apply.