Stat Assistant Professor

In what country did you do your BSc and MSc? It is very competitive for international applicants, and departments are heavily biased in favor of a few schools that have reputation of producing successful PhD students (Peking University, ISI, Tingshua to name a few). That said, it is not impossible either -- there are a few students in my current program from a few other places like Russia, Iran, South Korea, and Turkey, but most international are definitely from India and China. And the bar is set higher for international students too. Given this, your list of schools might be a bit of a reach if your grades were not solid A's, and you might want to aim lower and/or apply to do ANOTHER Masters program in the U.S. first and then go from that to a PhD (this is not unheard of -- many international students have had to redo a Masters in the U.S. and then proceed onto PhD). Your letters of recommendation should all be from professors. I think one applicant on this forum got into Harvard with one recommendation letter from a boss, but it was a research position. If you haven't been working in statistics research, all the letters need to be from professors. In stats, you do not need to contact professors beforehand, since incoming students are not paired with advisors like they are in other fields (e.g. where you have to be accepted to a professor's lab or research group). In stats, no one comes in with a doctoral thesis advisor. You can contact them if you want, but most won't think much of it or use the correspondence to recommend you for admission to the graduate admissions committee.

I think you have a good chance of getting into several of those. Berkeley seems to be a bit more biased towards those with a huge amount of mathematical preparation (so those with more math classes will be looked at more favorably), but even there, I think your chances there are not bad. I think you have a good shot at a lot of the others. For reference, an assistant professor in statistics at my Masters institution had a Bachelor's in Electrical Engineering and not a lot of math/stat classes beyond what you have, but she received her PhD in Stat at UW. It also seems to be more and more common for a lot of students to enter PhD programs in statistics without having taken real analysis too (so real analysis is now part of the PhD curriculum in stat at some schools, including the program I am attending now). As for recommendation letters, I would go with the lesser-known professor who can write you a *great* LOR. If you have two strong letters and one decent letter I wouldn't be surprised if you get into most of the places on that list.

I think that is a good list of schools to target. If you have strong letters of recommendation and can score decently on the subject GRE ("decent" typically means above 50th percentile, but you'd need 80+ percentile at the top programs like Berkeley and MIT -- but since you aren't aiming that high, I think 60+ would look good), I think your chances of getting into at least one of them are good. Adcoms will not place much weight on your honors thesis, or really, REUs for that matter (I have found from speaking with professors and seeing faculty comments on Stack Exchange that they tend not to care as much about these as letters of recommendation and grades in upper division math courses). So make sure you can get strong LORs.

A lot of MS programs in statistics in the U.S. are not competitive. Only a few are really competitive to get into (only Duke and Washington come to mind). I don't think the GPA issue is impossible to overcome, but in that case, you will need to have your letters of recommendation make a strong case that your GPA is not reflective of your ability. Also, is there any way you could do really well in your second year and THEN apply, or possibly retake some core statistics classes? Having a transcript that shows a second year of high grades and better grades on the second try may be enough to alleviate program concerns about you not being able to hack the coursework here. I realize that would delay applications by a year but in your case, that might be necessary if you *really* want to embark on the statistics path.

FWIW, at the program I am at now, there are several students who took no stats classes before entering the PhD program in stats who had Masters degrees in pure mathematics. I don't think lack of stats courses should really be a problem for the schools that you are targetting. For most stat PhD programs, it is better to have a lot of mathematical preparation than it is to have undergrad stats classes. (my school is also ranked in that vicinity of schools you have listed) That being said, you have barely just started your current PhD program! Just make sure you have thought it through very carefully before jumping ship. If your goal is to get a job in industry, a degree in mathematics is usually acceptable too, provided you have some computing experience and statistical knowledge. I know several folks with PhDs in math who now work in data science jobs that stat PhD graduates also find work in.

Ohio State is another possibility for stats. I really do think that for you, Stanford, Chicago, and Duke are unrealistic and that Berkeley is a bit of a reach, but you could try your luck at a few reach schools (doesn't hurt to try a few, if you have the money for the application fee). I would also apply to several Masters programs as well. If you do well there, it would be a good stepping stone to a PhD program.

For stats: Stanford, Duke, UChicago might be a bit of a stretch because of the overall GPA. Berkeley and NC State seem to be a bit more forgiving of lower overall GPAs (Berkeley less so though), so those are good places to aim for. I would suggest Carnegie Mellon, Florida, and Rice University as places to apply for stats too. These places seem to weigh recommendation letters more.

I am sure all of this helps, but for you, it may also depend on the reputation of your institution. It seems like American PhD programs particularly like to draw students from ISI (and even then, Masters students from ISI have a greater advantage than just those with a BS). I am not sure what other schools have an international reputation in statistics in India. One way to help get your foot in the door for American PhD programs is to do a Masters degree in mathematics and/or statistics in the U.S. They are less competitive to get into, so I would apply to a few Masters programs too if I were you.

I would not mention that you are reapplying (so I would not write "I previously unsuccesfully applied to PhD programs"), but you can draw attention to some of your Masters-level work. I would not mention the courses too much, since the adcoms can see them on your transcript. Instead, you could talk about some of the projects you have done and mention a few areas of statistics that interest you. Keep in relatively general though. My initial SoP was a bit too specific, but a more "general" one (where I mentioned interest in Bayesian methodology broadly and learning algorithms) I used seemed to be more well-received by schools (i.e. I got accepted at the schools where I used the "general" SoP).

I agree with footballman2399. I think it is likely you will get into at least a few top schools. If you remind your professors of the work you did in their classes (e.g. any projects or papers you wrote) and can get them to say something memorable about you, you'll be in good shape.

For mathematics, adcoms don't seem to care that much about REUs, since they rarely resemble actual PhD research. You may want to consider taking algebra in the fall quarter so you can get an algebra grade on your transcript before applications. The main things that adcoms are going to look at are grades in abstract algebra, real analysis, advanced linear algebra, and complex analysis (bonus points if you have topology, geometry, and some graduate courses on there), breadth of math courses taken, letters of recommendation, and subject GRE score. Given that you're coming from a top university, I think there might be less doubt about your ability to do graduate-level coursework, so algebra might be a better choice than graduate real analysis. Just something to consider... this is only my personal view though, maybe your advisor will know better and have different thoughts/better guidance.

UNC-Chapel Hill has a machine learning track in their INSTORE program. Purdue and Ohio State Stats departments also have research groups for statistical learning. I'd also take a look at schools you think you have reasonable shots at and see if there are any faculty working in the area of machine learning. If there are a decent number of them, I'd apply to those schools.

I disagree with TakeruK. I actually do think that this could be a problem, since abstract algebra and real analysis (and to lesser extent, topology, complex analysis, and geometry) are considered the "core" classes that pure math PhD programs (at least the top ones) are going to look at. The OP may be able to get away with not having an algebra class for an applicatoin to a lower ranked school, but not at the ones that s/he is targetting. Seing as the OP states that they would not have an algebra grade until March and most of the admission decisions will be made before then, this could certainly be a problem. However, there may be some other mitigating factors (like high GRE subject score and other graduate-level course work, strong letters, etc.). OP: Do you have any other graduate courses on your transcript? If not, one of the potential barriers I foresee is that the most successful applicants to top 20 pure math programs typically have a few graduate courses already under their belt. At MIT, Harvard, etc., it is not uncommon for the admitted candidates to have been taking grad-level math classes since their freshman or sophomore year of college.

I'm taking a 2-week road trip through a bunch of states and then I'll spend a few weeks moving in and getting acquainted with the area.

You could apply to both math and stat departments if your interests are in probability. I vote for PDEs (there is some cool research going on now about stochastic differential equations and stochastic PDEs). I'd also recommend a graduate-level one year probability-statistics sequence (if you have not already taken it).

For those classes, you will probably use Octave, MATLAB, and/or R for implementing AI and machine learning algorithms. If you take a statistical computing class, you'll like work with R, which is similar to Octave and MATLAB. I don't think you really need to take a stand-alone class on these. As long as you have a basic understanding of programming constructs (like if-then statements, for and while loops, assignment, and functions/passing variables <-- this is the big one) and basic matrix/vector algebra, you should be fine with it. Plus, there are TONS of online resources if you need to look up how to use a specific math function (like computing the inverse of a matrix, say), and there's no real reason to memorize these. You might need to learn Java and/or Python on your own, but again, there are so many online resources that you can teach yourself. It seems like data science and macine learing uses mainly these languages and/or R and Octave.

A large number of the "trickier" stats problems I've encountered involve completing the square, integration by parts, change of variables, and/or u-substitution. It'd be a good idea to know these thoroughly well.

Yes, but data science is more than just big data. You should take machine learning, statistical machine learning, and an artificial intelligence class. I think that, plus stats, should be sufficient for coursework. You should learn Hadoop, Java, Python, and MySQL on your own, and you should pursue internships. The internship experience will likely matter more than the courses.

I would ask your professors. It's likely that they have connections and collaborators at other universities, and having that "in" definitely helps when you apply. Given that you have a proven academic track record at a top undergrad institution and that you're now working at an elite one, I think you will likely get into at least some top programs. Good luck!

I think machine learning and data science are good bets though. I have seen positions starting at $90k at start-ups in places like Houston and Austin. The work seems pretty interesting as well -- ultimately, I think that is more important than salary. Commanding a very high salary may not be worth it if it is going to make you miserable!

You seem like a very strong candidate, but your list of schools is a bit top-heavy. Keep in mind that even people with close to 4.0 GPAs get rejected from these places. I would apply to some schools ranked below that too to maximize your chances of getting in somewhere. Best of luck.

These are all pretty lucrative. However, the jobs with the highest overall salaries are often located in expensive cities too. I have friends in the Bay area who are making six figure salaries and still have little disposable income leftover after paying for everything (and are also priced out of buying property).

A lot of possibilities. Biostatistics, data science, machine learning, actuary (if you pass the P and the FM exams), SAS or R programmer, etc. Try to get an internship in an area that interests you. That seems to help many grads land jobs a lot more quickly.

Both your major and your overall GPAs are very high. I wouldn't stress over the English class too much, but try to get at least a B-. Some applied math programs require the subject test, but a number of them also do not. There is often much more leniency on this than in pure math programs (e.g. any score above 50th percentile -- roughly the mid-600s -- is considered a good score for a lot of applied math programs, whereas many of the higher-ranked pure math programs want to see mid-to-high 700s and above). However, at the schools you listed, even the applied math applicants should have at least an 800 on the subject test (at least at MIT and Berkeley -- all applicants are expected to have achieved 800+ on it, and I know that they screen out the vast majority of applicants who score lower than that).

Yes, you should take Calculus-based probability and statistics. If you intend to apply for PhD programs eventually, it may be a good idea to take advanced Calculus/real analysis too. Beyond that, there isn't much that is "required," though having more math and stat courses signifies strong mathematical ability. Some additional classes that might be useful if they interest you: numerical analysis, ordinary differential equations, linear regression, design of experiments.