MathStat

I know this has been asked before a little while ago, but I just wanted to gather some more opinions about the new M1 Macbooks. I need to ditch my current Macbook Pro 2018 due to the never-ending keyboard and display issues. So I am looking to see what's out there, specifically for deep learning work. I've been using Tensorflow a lot lately, but I anticipate I will at least need to run / play with PyTorch code in the future too, with PyTorch being so popular in the academic community. I know that the integration of Tensorflow and PyTorch for Macbook M1's has been iffy, but for those who are familiar with this, have these issues been resolved lately (at least to a reasonably satisfying level)? What are some common glitches / library crashes that occur? I've already pulled up numerous guides explaining how to set up these laptops for deep learning work, but I just want to be aware what I'm getting myself into, before I take the plunge. Also, how do the Python basics like pandas and numpy work on the M1 (I've been finding contradicting info online)? and is it true that the PyCharm IDE works slower than before? Thank you very much in advance. I know I asked a lot of questions, but even partial replies would be very helpful for me. And if you guys have any recommendation for non-Mac machines, please do let me know. Thanks a bunch again!

I know so many people at top programs who took 1, 2, 3 or more gap years before starting grad school that I lost count. Some of these people spend their gap years doing either nothing or working underqualified jobs such as barista, crepe chef, etc. Don't worry about it.

You are on good hands at Duke, best of luck.

@stat_guy The "Wharton/Columbia/UChicago/Princeton" placements at Yale come from only 2 professors, one of which is Andrew Barron, a potential advisor I had considered myself when deciding grad schools. You should determine very carefully how active these two professors are currently, how motivated they are, and whether they are willing to advise you for the entire duration of your PhD (i.e. won't retire in a few years). As I mentioned above, Yale does have J Lafferty, a point brought up all the time both in favor of Yale and as a drawback to Chicago ("Chicago lost Lafferty, so it lost some of its value"), but is Lafferty actually actively advising students nowadays? Furthermore, many of the good placements came from David Pollard, but I think he must have retired by now. Having 2 senior profs agree to advise you before you start is indeed very nice, so I completely understand your inclination towards Yale. If you are sure you want to do mathematical statistics, then go ahead. Chicago would indeed offer more diverse research options, including hot areas such as ML and, more recently, causal inference. For math stat we have Chao Gao who is recognized as a clear outstanding researcher and who likely will become tenured soon (definitely during your PhD time). Tengyuan Liang form Booth is also an excellent choice and he will also likely become Assoc. Professor soon enough. Chicago may not promise you a famous advisor from the start, but you can certainly have one later on, assuming there is a match in research interests. Your first year, you could work with a motivated younger professor, then potentially add a famous advisor later on. I encourage you to come to the Chicago visit day on Monday and ask us questions. I can answer questions through PM, as well.

Is John Lafferty still advising?

I think rather than discussing program rankings we should make an advisors ranking, lol, but I guess that's too delicate of a topic even for an anonymous forum.

Is MIT access really feasible as a harvard stats student? If so then yup, huge point taken.

Bryon Aragam from Chicago Booth did UCLA stats and I think he is a pretty great positive example. When I was applying 2 years ago, I also read bad things about UCLA on here as well....

Chicago over Harvard was a no-brainer given my research interests. Of course that Harvard is still an outstanding program, and turning it down is not easy either way. Chicago over Berkeley was a hard and excruciating decision, but I reasoned that there were only a handful of Berkeley professors I truly wanted to work with all of which are absolute superstars and who have millions of students and postdocs...Chicago was an equally good match for my interests, they have TTIC which is at least top 5 for theoretical machine learning, as well as a handful of superstars or rising stars in the Statistics department and the Booth school of business. Also, besides purely academic reasons, between very long winters and very high living costs, I decided I'd prefer the former, haha. But this is a purely personal preference, and I totally understand people who think otherwise. You should also consider Chicago and should come at the visit! We do have the notorious quals, as well as the coursework which take up all of the first year plus summer. This is not for everyone. Research-wise, I am still recovering from that, yet I do have two exciting lines of research going on...I guess I just need a few more months to be able to tell you exactly how it's gonna turn out.

Are Harvard grads getting jobs at Google, Microsoft, Facebook, etc Research? I have not been stalking recently lol, but I cannot recall any examples (I wouldn't mind seeing some if you know any!). And the ones who get the top academic placements (aka berkeley and stanford TT professorships) seem to have worked in causal inference. If that is a strong area of your interests, then sure, go ahead. I turned down Harvard (and had similar interests to yours and even a similar situation, haha), cause I thought there were only 1-2 people with similar interests as mine. another bit of advice when you have to decide between almost all the top programs is to not get hung up on the top 2 ones that are ranked just after stanford. I think other ones you should also consider carefully given your interests are Columbia, UPenn Wharton, Duke, Yale (only if you wanna do pure math stat; imo they're some of the best at that). Make sure you review these carefully as well.

how is harvard a good fit given your research interests? I feel like they're more into biostat/applied stuff..although Cynthia Dwork is there... If you're into probability, deep learning etc, then berkeley and potentially other schools out of those 24 would be better fits.

Hi everyone, I would like to re-ask a question that I may have asked before in some comments. Please indulge me in this repetition. I would be very grateful for fresh/updated perspectives, too, especially from current professors (@StatAssistantProf @cyberwulf). I'm a student in my second year at a top 5 (or top 3?) stat PhD program (according to US news, but who knows). I would like to decide whether I'll pursue the industry or academia path by the end of the year. Since I've always liked theorems and proofs more than anything else, I may have an academic predilection. Now for the issue: I would really appreciate some candid/brutally honest comments on being the first student of a young assistant professor, but who certainly is a rising star. Of course, we all know about the tradeoff between rockstar/influential advisors who are incredibly busy and younger, enthusiastic, friendly, hand-on professors, yet who may not be able to help you that much on the recommendation letters front. Assuming the latter person is what I consider the best fit for me here as an advisor, then is it even worth bothering to pursue academia? Is there a way to feasibly make this work (any examples that you know of) or should I just forget about it, work on something cool for the next 3 years, then move on to something else? Thank you very much.

@stemstudent12345 Sorry you're going through that. A bit of pragmatic advice: Just find a prof or two who support you, that's all it takes; cram the patterns of past quals. Not worth wasting your mental health on this, *really*. You'll be done with quals soon enough and you'll be able to pursue your own exciting and beautiful research, and from then on nothing else will matter. @Stat Assistant Professoris spot on.

I am also starting to discover the connections between probability -> statistical physics -> spin model -> quantum field theory (arrows not necessarily in that order:P). Chatterjee at Stanford is more of a mathematician in my eyes, not really a statistician. But yes, interesting to think of the side of stat that comes from physics. And be able to use it to approach ML from a different perspective.

Basically the title. I know many prestigious finance firms ask for GPA, including major GPA, and sometimes transcripts, even for PhD students. Is that also true for tech companies? Especially for, say, competitive roles such as Google or FB data science? I know many people say grades do not matter anymore after you have passed your qualifying exams and advanced to candidacy, but I would like to double check this, perhaps with people here who have actually been through the process of industry job applications. As a UChicago student, you can all understand why I am asking this ?. Thank you so much!!

oh wow....having your writers customize letters would be a deal breaker for them probably... Mine just sent the same letter everywhere. I sent them the request links all at once, so it only took them like 30 min to finish the whole process.

If you have no promises of acceptances (such as, for instance, at your alma mater; some people are lucky and they know their PI can take them on for PhD), I would apply to as many schools as you can afford. I applied to 20 which was overkill, but then again, I had no guarantees of acceptances, and was scared by an example of one guy here with a similar profile to mine who applied to 15+ and got into nowhere. I heard postdocs apply to 60+ positions, so we probably shouldn't complain about sending many Phd applications (besides I have a feeling nobody reads the SOP, they just look at your letters). Would love to hear how many applications it takes to crack a tenure track position.

Congrats @Bayequentist! I really wish I took the extra EE/CS classes in my first year (or self studied them, when they overlapped with the core statistics classes...), but I did not manage to achieve this, haha

I just finished the notorious first year of coursework plus preliminary exams at the UChicago Statistics PhD program. Happy to report that I am still alive, and dare I say, excited to move on to research, despite the new set of struggles and uncertainties that will come hand in hand. I do face the following issue now, which seems to be pretty common within US Statistics PhD programs - I recall, for instance, this very heated discussion from a few weeks ago, which did resonate with me a lot: https://forum.thegradcafe.com/topic/125581-school-suggestions/?tab=comments#comment-1058776870. Similarly to that post, I also felt that the first year coursework focuses on traditional statistics topics (in my case, linear regression, GLMs, overview of Bayesian methods, old-fashioned math stat, such as complete, sufficient, etc. statistics, UMVUE, minimaxity, admissibility, James-Stein estimators, and then some modern math stat, such as EM and variational Bayes algorithms, regularization methods, hypothesis testing and multiple comparisons), yet it misses some core courses needed for those who want to do modern ML research (guess I joined the dark side, too, despite starting off as a probabilist with interests in mathematical statistics). The particular courses I have in mind are optimization and statistical learning theory (and who knows what else I'm missing). I am now trying to address this by taking a very strong learning theory course, yet I do not have time to wait for the optimization course which will be offered later. So as silly as this sounds, my question is, how does one efficiently self study new material relevant for their research, especially while balancing other courses, research, TAing, etc...? I feel that in order to gain the most thorough and solid preparatio, one should take the past course materials and do all the grueling long homeworks, readings and so on, but then again, there are those other time constraints I just mentioned. I'd love to hear some advice from more experienced posters on how they pick up the skills needed for their research as they go. Thanks a lot!

Great response by @Stat Assistant Professor. Casella Berger is already assumed knowledge for some top programs. But if you are admitted based on your pure math background (like yours truly) you likely won't have even cracked open Casella Berger or have taken a proper mathematical statistics course before coming in. However (and I hope this is not too strong of an opinion), Casella Berger presents math stat in a really outdated way. More modern and useful texts nowadays are from Van der Vaart, Lehmann and Casella, Iain Johnstone's Gaussian sequence model book, etc.. IMO Stanford does their math stat sequence in the best and most modern way. Their lectures and homeworks are online. The last part of their sequence, STAT 300C focuses on multiple testing, which is a very hot topic nowadays. Also, "hardness" of a program is a really subjective thing. We only discuss about the coursework and preliminary exam requirements above. For me, what also constitutes a big chunk of "hardness" is whether you'll be able to find a strong advisor that you like, whether you'll be allowed to start research asap etc, etc.

I could definitely see you get into either duke math or stat phd with your profile. I'd apply to duke math if I were you (several profs in the duke math department have cross appointments with stat, eecs, etc.; many of them do great ML research; the department is also extremely supportive; the math phd cohorts are a little smaller than the stats ones; however the stat program is also great). And I agree with @StatsG0d, your lists are somewhat bottom heavy.

i've seen many schools waived their Math gre requirement on their website.

First of all don't feel bad about a bad start, it happens. I have gotten to learn many stories of famous statisticians who partied in undergrad then discovered their math/stat passion and made it to top programs and very successful careers later on. From my experience with Stat PhD admissions, a sure-fire way to earn the admission committee's respect is to get As in grad level math classes, particularly in analysis and probability (which build the foundation of statistics). If I were you, I'd work my way up by taking the honors analysis classes offered at your school, then measure theory, then measure theoretical probability. Taking the courses I mentioned should take 2 years. In addition to that, a sure-fire way to impress committees is to have math research with professors at your university. In all honestly, what I have noticed is that, at many universities, math professors are much more open to taking students for research projects than stat professors. Of course, stat professors can also mentor you, but typically, undergrad stat research is just doing a small improvement to an algorithm or working with data which IMO is not enough to shine in your applications. All of this should take your junior and senior year, then perhaps you should take a gap year in which you work as a research assistant/in industry and in which you complete your actual applications to programs. In all honesty, I think masters programs entail such a massive financial burden and/or debts that I am seriously wondering if they are worth it anymore. At the same time, going to a very poorly ranked/bad department is not really ideal, although if there are good advisors there/good industry placements, then I guess, why not? These are the main reasons why I suggested a gap year to gain more time and build up a strong app for top places. By doing this you will at the same time figure out if a stat phd is what you really want (I think you have not yet taken enough courses to form a strong opinion, but you should figure out quickly once you take more advanced classes.) Of course, take all this advice with a grain of salt as it is partially based on the path I took (i.e. math grad level classes and math research) which yielded much better results than I expected. It also tremendously helped me to have letters from famous professors, so try to work with such people at your university. There are of course many more possible paths, and I will let more experience posters on this forum comment on this too.

@cyberwulf I know someone who *submitted* a paper to one of the top 2 stat journals (Annals/JASA) (so not published but still extremely impressive). However, I assume it would be hard to have an actual publication before applying to the PhD since that would mean you would need to submit at least very early in your junior year (assuming you apply straight out of undergrad), which means you need to start the research at least sometime in your sophomore year. ---- Btw, how long does the review process for Annals of JASA take usually?

Yeah your grades would matter a lot. Definitely for US PhD applications, as they care a lot about your ability to handle grad level courses, just like they care when you apply as an undergrad (you're essentially starting over your application progress, that's how "transferring" in the US works). For Europe PhDs I don't know for sure, but I am somewhat familiar with the UK system (applied there for both undergrad and grad but chose the US both times) and the UK is very grade-centric, they seem to value grades more than research/letters.