StatsG0d

I'm actually taking the online Probability Theory course via Penn State World Campus now and I think it's pretty great. I thought that it would be easy and applied (even though it says theory) but we actually do rigorous proofs in the course. I would recommend it.

I think you'd be competitive regardless of whether many people apply or not, but I doubt that any masters at Columbia will be non-competitive even if the program is expensive. You might as well take a shot at applying. The marginal benefit far exceeds the marginal cost (of applying).

I think it'll be tough to get into those schools with that GPA and in particular the C+ in analysis especially since your school is not highly ranked. Take what I say with a grain of salt as I'm applying in the same cohort as you. The rest of your profile looks great though. I don't see CMU as that far of a reach but UWashington I think would be very tough. Then again, you have a coauthored paper so that helps you out a lot and your GRE scores are high. I think one strategy might be for you to take the math subject GRE to show that although your math GPA isn't the highest amongst the top applicants your math skills exceed them. Your advantage is that you have taken many math courses, so you can probably do well on it if you study hard.

You can also get Federal Work Study. I'm not sure, but maybe the FWS can be used to subsidize TA or RA duties, so that funding you is slightly less costly?

164 is 88th percentile for quant. It's 93rd for verbal.

I agree about the GPA, but I guess if you're coming from a program known for grade deflation like U Chicago, it might be overlooked. Otherwise, that's a tough weight against you.

That's true and I've read a lot about that--so I guess I have a decent shot. My other concern was being from a relatively unknown low-ranked university. I'm just hoping that my decent grades in math courses and my research assistant experience can help offset that.

From what I have read, a lot of people are coming from industry (including me, but I worked for an international organization as opposed to a private company). I doubt it would hurt you, and if anything it could help you if you can show off your new skills and how the experience shaped you to be a researcher.

I see--so it's actually more of a supply issue than a demand issue. Any suggestions on programs, Applied Math to Stat? Also Statisticsfall2014, I believe I was following your advice on your blog in regards to applying to the NSF. I found it extremely useful thus far. Thanks for that.

Thanks a lot. I guess my biggest concern is my mathematical preparation and somewhat ethnicity. I was browsing through Purdue's grad students and almost all were Asian and were math majors (presumably taking many more math courses than me). I'll definitely consider those two schools. Thanks again.

Hey everyone, Several months ago, I made the possibility of having to retake the GRE (I have decided now not to do that). Now that the application process has begun (and I have been hard at work on applying for the NSF GRFP), I am trying to decide on which schools to apply to. I have various interests, including (in no specific order) machine learning, financial, and bayesian statistics. I have a geographical preference for the midwest (in particular around Chicago), but it is not a strict requirement. I was thinking of applying to the following schools: Purdue UIUC Iowa Michigan St. Ohio St. Florida U Michigan U Chicago (MS) Boston U NYU (Stern) UCSB UCSD Duke Texas A&M Washington (far far reach) The main school I would like to go to is Purdue because the department seems to be large and have faculty in several areas of interest. Given that I am not that specific yet about what I want to do in statistics, I think a good strategy would be to choose such a department. Obviously, an ivy league school would be great but it just is not realistic. Any advice people could give would be appreciated. Undergrad Institution: Top 100 Private Program: BA Economics, Minor in Mathematics GPA: 3.81 (>3.9 excluding study abroad) Type of Student: Domestic White Male GRE: 164Q / 159V / 5.0 Quantitative Coursework: Mathematics: Calculus I (B+) Calculus II (A) Calculus III (B+) Differential Equations (A) Linear Algebra (A-) Number Theory (A-) Real Analysis (A). Statistics: Intro to Statistics [calc-based] (A), SAS Programming (A) Econometrics (A). Currently taking probability theory online via Penn State, and doing well. Computer Science: Object Oriented Programming (A) Intro to Computing (A-) Research/Work Experience: At the time of application, 1.5 years research assistant experience at an international organization in Washington, DC. Did things like data collecting, regressions, Stata programming, and also listed as a main contributor on a popular publication. Will have published 1-2 working papers (likely not to be in a journal, but on the web site). 1 year research experience with an economics professor at my undergrad university. Highly regarded (top 5% in IDEAS for economics). Programming Languages: Advanced: Stata Intermediate: SAS, Python, C# Learning: R Awards/Honors/Recognitions: Magna Cum Laude, Alpha Sigma Nu Honor Society Letters of Recommendation: One from my boss, my former boss for whom I was an RA and also a student, and my former statistics professor, who is well regarded in Biostatistics (all cited several hundred times).

It would probably he helpful if you posted which math courses you've taken as well as your grades in them. Also, you may want to list your research interests. That said, your undergrad GPA is quite low and I think a masters does not directly replace undergrad performance, particularly if the masters was completed outside the US. Cyberwulf has often mentioned that the competition for international students is very keen, and his department only accepts students from international schools that have a track record for producing successful phd students. I suppose if you went to one of those, you have a decent shot.

Test anxiety. Trust me I took it twice. First time studied for 3 weeks and got a 163. Then studied for about 5 months and got a 164 (lol). The rest of your application looks solid. If you really don't want to take it again, you could try taking the math subject test. If you do really well (>= 85th percentile) then you probably have a decent shot of cracking the top 15 even with the low general test score.

Purdue has a joint CS-statistics program.

I wouldn't say that they have nothing in common. The first year PhD sequence in finance is almost identical of that to econ. Econ tends to open up more opportunities than finance, but finance tends to pay better than econ. Most would recommend getting a PhD in economics purely for the job opportunities associated with it. The consensus seems to be that a PhD in economics can be a professor of finance, but a PhD in finance cannot be a professor in economics.

Have you taken multivariable calculus? If not, I think that your chances would be unlikely. Also, you need some exposure to a proof-based class like real analysis.

I see your point, cyberwulf, and thanks for the advice. Perhaps my proposal is better suited for a NSF GRFP than for an SOP.

One area of interest I had in particular was bridging the gap between econometrics and machine learning. More specifically, I would like to apply econometric techniques such as instrumental variables and quasi experiments to machine learning analysis for causal interpretation. The only problem is that I feel this may be too specific to write about and that the subject itself is presumably audacious. Any thoughts?

You should definitely retake it if you're set on doing CS grad school (or any kind of quantitative grad school really). Calc II is usually a prerequisite and I'm sure the admissions committees expect a strong grade in the class. You'll also need multivariable calculus and likely will need to get at least a C+ in order to advance onto that class. Good luck.

I'm in a similar boat as you as ML is one of my research interests. Indeed, it does seem to be hard to find programs that specialize in ML, probably because most ML researchers (as far as I have seen) are housed in computer science departments. Many of these, however, have joint appointments in CS and stats departments. Duke looks to be an additional choice, which is highly ranked but at least not as high as Chicago / CMU / WashU / Wharton UCLA offers ML as a subfield.

Again, take my advice with a grain of salt since I have not gone through the process (we will be in the same cohort), but I have read a lot from this forum and elsewhere. From my understanding, the adcoms take a holistic view at the application to determine research potential. The letters are especially crucial if (as in the case of yours truly) you have less than spectacular grades or test scores, or are coming from a relatively unknown university. Given that this is not the case for you, I don't think it would be absolutely vital for you to have amazing letters, as long as they are completely positive in tone (e.g., they say that you would make a poor researcher). I recall one poster who graduated from ivy league with a top GPA and top scores and got rejected across the board, which cyberwulf and biostat_prof appeared to attribute to his being an international applicant (not sure which you are). The general advice is apply broadly. I would be truly surprised if you did not make it into at least one school in the top 15, let alone the top 10.

You have a near perfect GPA as a math major from an ivy league school and an awesome GRE score for both the subject and general tests. I would say this alone would make you a strong candidate pretty much anywhere. Given your grades in math classes, why would you not have strong letters from your math professors? Take this with a grain of salt as I will be applying to the same cohort as you, but I would say that a letter from a humanities professor, despite how strong it is, will not necessarily weight in your favor since he likely won't be able to comment on your mathematical strength. I would say get letters from professors who can attest to your quantitative ability. Just my .02. That said, your profile is otherwise very impressive, more so than many of the profiles I have seen on here that made the top 10. See if you can get involved on some RA work and I would bet my hat you could get into at least a few of the top 10.

As an update, I got a 5.0 on the writing (but I don't think this matters much for statistics?)

That's pretty reassuring clurp. Thanks a lot, and congrats to you.

Take this with a grain of salt, but from my understanding AI and ML courses are more mathematically-centric than computer science-centric. Given your background, you're likely to be successful at pretty much anything you decide. It seems like the hardest thing for you is deciding. My background is in economics (with a math minor). After I took my first econ course, I thought 100% that I wanted to be an economist. After working as an RA for the last 2.5 years for economists, I decided that it isn't for me. Empirical economists tend to rely on strong assumptions (e.g., rationality) and they tend to use one tool: linear regression. From my viewpoint, there are many relationships which are nonlinear, so I decided to go the statistics route. I think your REU will provide you with a great experience to see if you like quantitative finance. It's an intriguing field, and perhaps will be my preferred research area when I (perhaps I should say if I get accepted) go to grad school. Despite all that, from what I read, it seems like you would be a great mathematician. On the other hand, the best economists are mathematicians. The two need not be mutually exclusive, as virtually all theory of economics comes from mathematical relationships. To play devils advocate a bit, I think you would be successful as a econometrics researcher. You would be able to develop methodologies to advance econometrics. All these opportunities you describe sound fantastic. I wish you luck in your endeavors. Perhaps you should look for research opportunities with professors or with organizations such as the IMF, World Bank, and the Fed if you want to learn more about economics research.