Jump to content

Bayequentist

Members
  • Posts

    37
  • Joined

  • Last visited

Reputation Activity

  1. Like
    Bayequentist got a reaction from spartans in UChicago vs CMU: Where would you go for a statistics PhD?   
    I would take the results of the poll with a grain of salt because this question is heavily based on personal preference/ability.
  2. Upvote
    Bayequentist reacted to DanielWarlock in UChicago vs CMU: Where would you go for a statistics PhD?   
    Both schools are focused around the theme of high-dimensional stats. But risking oversimplification, a quick summary of their difference is: CMU is more "CS"; UChicago is more "mathematical". If you consider yourself more of a mathematician/probabilist, go to Chicago. If you consider yourself a computer scientist who looks at more applied stuff, then go to CMU. I will now further explain what I mean. 
    CMU focuses more heavily on more applied, interdisciplinary stuff like neurosciences, astrostatistics, social sciences and yes sports analytics.  Of course, most of these are done under the tag of "high-dimensional statistics".  But I would even go so far to say CMU stats has more of a "CS flavour" if you know what I mean. On a related note, CMU is also much stronger on causal inference, which is also more "CS". I also feel the organization is very similar to what I see at EECS at MIT: they have all these themed working groups like "astrostats group", "causal inference group". So the community based activities like colloquium/talks, reading groups will be more specifically tailored to your subfield. 
    Chicago is more theoretical and will probably be more so in the coming years based on their new hires. Maybe "theoretical" is not a good descriptor. What I mean is that their new hires now mostly have tags such as "physics", "statistical mechanics", "random matrices", "Fourier/harmonic analysis", "combinatorics", "random graphs". Chicago is definitely more mathematical and has a taste of more probabilistic things. They even hired student of Borodin who does hard-core math. 
  3. Upvote
    Bayequentist got a reaction from statenth in What to consider in choosing a phd offer   
    The points mentioned by @BL4CKxP3NGU1N are all good. Additionally, you should also find out if your potential PhD university has a strong grad student union or not. Having a strong grad student union comes with many benefits like good health insurance, emergency fund, annual contract bargaining with the university etc... These "small" things make your life as a grad student (potentially 5+ years) much easier.
  4. Like
    Bayequentist reacted to DanielWarlock in Most efficient way to self study material required for research   
    To master a technique for me is very very hard. In fact, I often find that taking even a very solid class does not truly allow me to master a technique--in the sense that I can independently solve a problem using that technique. To give an example, I first learned the gaussian interpolation in a class in the context of Slepian lemma. Then I read Vershynin's book and learned it again, this time not only Slepian but also its extension such as Gordon's inequality. I even derived Gordon's inequality using interpolation as an exercise from the book.
    Now when I see it again in the context of spin glass (Guerra's work on existence of free energy and upper bound), I stumbled as a total novice. I tried to prove these two theorems on my own without looking at the proof. Again, it proves to be quite a challenge and I just can't do it. So I studied interpolation fourth and fifth times. Later the monograph poses an exercise to use interpolation--again this takes me hours to finally solve on my own. You could imagine that to apply interpolation in a research problem in a nontrivial way could be much more challenging. So I still have a long way to go before I can claim myself a master of interpolation.
    So in a sense, taking a class is as quick and efficient as it can get but in a way I also feel it is less "nutritious" a bit like junk food. Many classes (at my institution at least) feel like a guided tour around an amusement park. You see "prototypical arguments" of a lot of stuff in its simplest form, but it never gives you a feeling that you are "hitting it hard enough" by working out all the different variants. 
     
     
     
  5. Upvote
    Bayequentist reacted to Stat Assistant Professor in Most efficient way to self study material required for research   
    I often find that the best way to learn a new field/subject is to watch video lectures, read review articles and read select chapters from textbooks. So when I wanted to learn about variational inference, the first thing I did was watch a few video tutorials by David Blei and Tamara Broderick. After establishing this "baseline," I kind of just pick up on things as I go -- i.e. I just read the papers and try to figure out what the authors are doing as I go. This gets easier to do as you gain more experience and as you read more papers (in the beginning, I might annotate the papers a lot more). 
    Realistically, when you are doing research, you won't know (or need to know) *everything* there is to know about, say, convex or nonconvex optimization. But you can pick up what it is you need as you go, and if you encounter something you're not familiar with, you get better at knowing WHERE to look and fill in those gaps. 
  6. Upvote
    Bayequentist reacted to MathStat in Most efficient way to self study material required for research   
    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!
     
  7. Upvote
    Bayequentist reacted to bayessays in USWNR Statistics Rankings   
    I believe they rank them every 4 years. The last ones just came out in 2018, so I'd expect new ones in 2022.  The biggest changes are the additions of new programs that were previously not ranked.
  8. Upvote
    Bayequentist reacted to bayessays in Masters in Statistics (low gpa)   
    Contacting professors, besides questions for the admissions chair, in a statistics department when applying for a MS degree makes absolutely no sense.  Someone with almost straight A's in math classes taking the math GRE, a test they will almost certainly not do well on, for absolutely no reason makes no sense.  I'm not arguing from any position of authority, so stop with the personal attacks and rude tone.  I'm trying to help someone and you're spreading misinformation. 
  9. Upvote
    Bayequentist got a reaction from Stat Assistant Professor in Masters in Statistics (low gpa)   
    Agree that reaching out might help, but it won't help most of the time. From the pinned post by cyberwulf: Funding in most (but not all) U.S. stat/biostat programs is allocated at the department level to the strongest incoming students, so applicants aren't typically "matched" to potential advisors who agree to fund them*. Rather, the department projects the total number of positions available and then tries to recruit up to that number of students. Once the students are on campus, they are then either assigned to a position or (ideally) have some choices available to them. Of course OP should still try and reach out to faculty (but don't expect anything).
    Regarding GRE subject test, OP did not take Abstract Algebra, Real and Complex Analysis, so taking the test will most likely mean throwing money away. Still, if OP is independently wealthy and willing to give it a shot then by all means go ahead and take the test.
  10. Upvote
    Bayequentist reacted to cyberwulf in Does extracurricular activities matter at all for PhD admissions?   
    Just include extracurriculars and don't worry about it. At worst they'll have no impact, and at best they might catch an admissions committee member's eye as a slight positive.
  11. Upvote
    Bayequentist reacted to Stat Assistant Professor in Opinions on stats programs that don't require advanced statistical theory or measure-theoretic probability?   
    I observed this even in the case for the first-year Masters-level Mathematical Statistics sequence at my PhD program. The first semester, based on chapters 1-5 of Casella & Berger, is more-or-less the same, but the second semester now deviates from Casella & Berger quite a bit. They used to spend a ton of time on things like UMVUE, Neyman-Pearson, and Karlin-Rudin, but now, they either skip it or abridge it considerably, and instead, focus on topics like the EM algorithm, lasso and ridge regression, etc. By now, things like EM algorithm and lasso are not that "new," but they're certainly not relatively archaic like UMVUE or UMP tests, and they will probably be standard tools used for awhile.
    I think it's a good thing. But then again, when I started to do research, I was basically learning everything on my own (I could go to my advisor for help and questions). So I can't say that most of the classes were really directly useful for research, but it didn't end up mattering in the end anyway.
  12. Upvote
    Bayequentist reacted to DanielWarlock in Opinions on stats programs that don't require advanced statistical theory or measure-theoretic probability?   
    I completely agree with doc's assessment. In fact, I can observe this trend at Harvard. The inference class this year is taught from a range of relatively modern topics instead of unnecessarily rigorous proofs on consistency and normality of MLE/UMVUE/NP tests and stuff like that. The measure theoretic has been downplayed a lot at Harvard as well because it is "almost completely useless". That said, the classic asymptotic techniques are still very useful. When you write some research paper, it is expected that you will give some bounding statements with *NO* exceptions and the toolkit/intuitions for doing that is pretty standard from the classics.
  13. Upvote
    Bayequentist reacted to Stat Assistant Professor in Opinions on stats programs that don't require advanced statistical theory or measure-theoretic probability?   
    What are the job placements like for the schools you mentioned? For industry, it probably makes no difference. For academia, having to take these courses may be helpful in that they allow you to sharpen your proof skills, and you pick up on certain techniques from them that you can use repeatedly in your research (splitting the expectation ftw). But if you read enough papers carefully, you can probably also pick up on "standard" proof techniques.
    For academic hiring at research universities, it's most important that your *research* is prolific and at least some of it is cutting-edge (i.e. getting published in the top journals or top machine learning conferences), not the content or grades of your coursework.  
    Anyway, my two cents: Lehman and Casella is a very classical text but a lot of the material in it may not be very relevant to most modern statistics research (for example, L&C gives a *very* rigorous treatment of UMP tests, admissible estimators/tests, etc., which isn't a popular research topic now). I guess it's nice in that L&C has a lot of material on things like James-Stein estimation that was one of the earliest shrinkage methods (before lasso and all the sparse regression methods). But is it really necessary to know the risk/minimaxity properties of these kinds of estimators in great detail? I'm not sure.
    As for probability theory, I definitely think it's good to be able to understand notation for the Lebesgue integral and know basic inequalities (e.g. union bound), but if you're a statistician and not a probabilitist, you may be able to get away with only the basics. I believe that at UC Berkeley, PhD students in Statistics do not even need to take measure-theoretic probability (they can instead take only the Applied Statistics and Theoretical Statistics sequence), and their PhD graduates seem to get along just fine.
  14. Upvote
    Bayequentist reacted to Stat Assistant Professor in Pursuing a PhD in Statistics & Data Science for professional reasons - overcoming feeling of inadequacy due to "passion"   
    I do know several Statistics PhD graduates who now work as Research Scientists at Facebook, Google Brain, and Amazon (some after spending a summer being a Research Scientist-Machine Learning intern). I don't really know what they do as far as "research" goes and/or if that is any different from "regular" data scientists, though. If you want to do *really* basic research like computing theory or mathematical foundations of statistical learning outside of academia, the opportunities will indeed be very limited (e.g. Microsoft Research, maybe some national labs).
    Anyway, I would figure out what your priorities in your work/non-work life are and go from there. Even I decided that I was not going to be a postdoc for more than three years and that I would go back to industry if I couldn't find an academic job within 3 years of finishing my PhD (fortunately I found one during the second year of my postdoc). If I had to leave academia, though, I would have definitely missed academic research but I think I would have been fine too -- if I didn't find work sufficiently satisfying/stimulating, I would have redirected that energy into making my non-work life satisfying (maybe even by taking some MOOC's to keep my brain active, likes bayessays suggested).
  15. Like
    Bayequentist got a reaction from sante951 in Are my math courses sufficient?   
    Besides classes, GRE General/Subject, it'd be awesome if you can do a REU this coming summer. A good LoR from a research advisor will strengthen your profile a lot.
  16. Like
    Bayequentist reacted to BL250604 in Profile Evaluation: MS/PhD in Bioinformatics, MS in Statistics   
    Absolutely agree with @Bayequentist, I think you'd have a very good shot of getting funded at Oregon State. Perhaps even South Carolina (they do fund M.S. students as GA's), UVA (not sure on their funding situation with M.S. students) and UMass Amherst are all decent spots that can set you up well into a better Ph.D. program or, if you like the program, you can stay there for the Ph.D. 
  17. Upvote
    Bayequentist reacted to aluc in Early assessment for a non-traditional applicant   
    Hi again,
    With deadlines getting closer I was hoping to get some advice on a more detailed list of schools. Two important updates that affect my application: I don't think the paper I am working on will be submitted by the earliest deadlines, and I scored a 164Q/167V on the GRE. Could anyone shed light on whether it is worth taking again? It seems like the general quantitative cutoff mentioned on this forum is 165. It'd be a pain to do it for the few point increase, but if the current score precludes me from even being considered by most schools on my list, then obviously there is no choice.
    I find it tough to label departments as reaches or safeties with any certainty, so I'll just list them roughly in the order of their rankings. I took @bayessays advice and am trying to cast a wide net in the 25-50 range, with a few lower ranked. There are a few places I'm not considering for geographic reasons, but just in case are there any obvious Bayesian departments I am missing, or any reaches I should outright remove to save myself the application fees?
    NCSU Iowa State UC Davis UCLA Ohio State Rutgers UC Irvine UT Austin UPitt Mizzou UCSB Oregon State UCSC Columbia Biostat Penn Biostat Minnesota Biostat Thanks again for any advice.
  18. Upvote
    Bayequentist got a reaction from bayessays in Course selection for prospective statistics applicant   
    In addition to numerical analysis/optimization, I think you should take a legit CS class, like Data Structures & Algorithms or Parallel Programming.
  19. Upvote
    Bayequentist reacted to bayessays in Course selection for prospective statistics applicant   
    Once you already havw linear algebra, real analysis/measure theory, mathematical probability and statistics, and perhaps some numerical analysis/optimization courses, it won't make a huge difference what exact courses you choose as electives. If the stats courses are completely applied where you don't do a lot of math, then I would lean towards math but you already have so much math that you'll be good even if you never take another class.
  20. Upvote
    Bayequentist reacted to bayessays in Best probability textbook for self-study?   
    Stat 110 is a typical undergraduate probability course - your first semester Casella Berger course in grad school is just a slightly harder version.  Mathematical statistics is the next chapters of the book - probability predicts how a model will generate data, mathematical statistics builds on those tools to go backwards, figuring out model parameters that fit given data.
    Anything more than reviewing basics of linear algebra, how to do a convergence proof in real analysis, and integration by parts will be overkill.
    Grad school is not about taking and passing classes - you are going to pass your classes unless you don't show up for the tests.  You will never know every piece of mathematics, so preparing for it by thinking you can just accumulate all the knowledge before it doesn't make sense.  I, like you, tried to learn all this stuff before grad school (and during grad school for future, and learning new programming languages I might need in future). It was all a waste of time.
    That being said, if you like reviewing math, go for it! But I don't think you need to be provided specific books because you already, this moment, know what you need to succeed or they wouldn't let you in.  If you feel the need to do something "productive" to prepare, spending that hour a day learning some Chinese to help make friends with half your future classmates, or getting into some kind of exercise routine BEFORE school starts, will do way more for you than learning this one weird linear algebra trick.
  21. Like
    Bayequentist reacted to Stat Assistant Professor in Elements of Statistical Learning (Hastie, Tibshirani, Friedman)   
    These are very nascent fields (I've heard that even computer science conferences have difficulty finding suitable reviewers for papers on theory for deep learining) and so are fruitful areas for research. But since they are so recent, deep learning is not covered in The Elements of Statistical Learning (to the best of my knowledge -- there may have been a more recent edition of it than the one I read that contains sections on deep learning).
    I would also note that ESL is by no means exhaustive. The authors are frequentist statisticians, so the book only very briefly touches upon Bayesian approaches to the same problems in the book (e.g. nonparametric regression, classification, etc.). But I think as a "gentle" introduction to a variety of machine learning methods and models, the book is quite good. IMO, this book is best utilized as a basic introduction and should be supplemented by reading other tutorials and lecture notes/slides and watching video tutorials.
  22. Upvote
    Bayequentist reacted to WornOutGrad in Housing and Grad school   
    I decided to live in the dorms my first year of Grad School. I went for the cheapest thing on campus...

    BIG MISTAKE!!!!!!!!!!!!!!

    I have a freshman (yes, an undergrad freshman) roommate who loves to stay up until 4, playing WOW (including the night before I had a midterm), and the building smells like weed most of the time. I don't think it's until now that I realized how immature underclassmen are! So if you stay on campus, splurge a little bit for a graduate apartment, because this 4am World of Warcraft crap and the weed have got to go!

  23. Like
    Bayequentist reacted to Stat Assistant Professor in Elements of Statistical Learning (Hastie, Tibshirani, Friedman)   
    "The Elements of Statistical Learning" is a good book and I found it to be pretty accessible. If you have already finished the first year of your PhD program, you probably can just start reading it (and then refer to other references when you need to). I think it's a good idea to read this book from start to finish at a high-level (i.e. read it enough to understand conceptually what's going on rather than focusing on the nitty gritty technical details). Then you can refer back to parts of it later and read those sections much more carefully if they are related to your research. I never did any of the problems at the end of the chapter, but I don't think that would have helped much for research anyway.
    The best way to "start out" research is not to dive deeply into theory and technical details, but to get a broad sense of the field and then narrow the scope of your project (your PhD advisor should guide you to picking a suitable project that is not too ambitious and not too incremental but "just right" and doable enough to result in a paper or two). Reading through this book, watching tutorial videos, and reading review articles, lecture slides, tutorials, etc. are all great ways to learn the basics/high-level background of statistical learning. Once you start research, your advisor will probably give you a few important/"seminal" papers to read as well, and then it will make sense to delve into technical details, theory, etc.
  24. Like
    Bayequentist reacted to Stat Assistant Professor in What major should i choose in Grad school for Machine learning   
    You can study machine learning in either a CS or a Statistics department -- indeed, there is a great deal of overlap between the two fields when it comes to machine learning, and a lot of the faculty will be cross-listed between Stat and CS/EE. That said, I would say there is a bit of a cultural difference in ML between Statistics and CS departments, in that CS people tend to be more focused on algorithmic aspects and predictive accuracy of point estimates (so models that work well in practice and that can be implemented with greater computational efficiency are preferred). Computer scientists seem to approach problems from more of an engineering perspective: how do we build a system that works to solve whatever problem we have? In contrast, Statistics people tend to be more interested in statistical inference (confidence/credible sets that quantify uncertainty of estimates) and theoretical properties of estimates like asymptotic/large-sample behavior, risk bounds, etc. There is of course still a great deal of overlap, with many computer scientists studying theory of machine learning and many statisticians becoming increasingly interested in scalability and computational complexity (which often involves borrowing tools from systems design, like distributed/parallel computing, etc.). But in general, it seems as though the underlying mathematical foundations and statistical properties are emphasized a bit more in Statistics departments, while the algorithmic aspects and empirical predictive quality are emphasized more in CS. 
    Additionally, the admissions process for CS is quite different from Statistics. For CS, strong research experience is basically a requirement for getting into a reputable PhD program, and a slightly lower GPA is often forgivable in PhD admissions if you have excellent research experience and worked on papers with well-known faculty. Admissions in Statistics programs focuses less on research experience (though it is a plus) and more on "hard" numbers like GPA, grades in upper division math classes, breadth of math courses taken, and GRE scores. "Research potential" is assessed more often by letters of recommendation in Statistics, and there are a lot of students accepted into Stat PhD programs who have no experience with undergraduate statistics and little research experience, but a lot of academic experience in pure mathematics.
  25. Like
    Bayequentist reacted to DualRoasting in Fall 2019 Statistics Applicant Thread   
    Turned down iowa state yesterday. Good luck to those still waiting!
×
×
  • Create New...

Important Information

This website uses cookies to ensure you get the best experience on our website. See our Privacy Policy and Terms of Use