fluffyj

I got a B- in a data analysis class, a B+ in algorithms, and a B in real analysis II all of which are much more damning than a B in computer architecture, but still got into a top 5 program and waitlisted at another. My major/overall GPA was very similar to yours. I wouldn't worry about it. It's more important to have an upward trend in grades and have other factors like research or letters of rec in your favor.

Damn congrats. Do you mind sharing your stats/application profile? As a future applicant, I'm curious to know. If you haven't already found this, Berkeley's stats program has a wiki run by students with detailed info about the program. I agree you can't go wrong with either choice and I've heard both schools are very strong for theoretical stats.

I think it looks good. Did you do any research in undergrad? Also, linear algebra is often listed as a required/recommended course for statistics phd programs (along with analysis) so that might be cause for concern, but I think it should be okay if it was a long time ago and if you took graduate algebra classes in your master's program (which should cover linear algebra if it was along the lines of Dummit and Foote or something similar). I don't think you should be too worried about the stats courses since it seems like you have taken much harder courses and gotten good grades in those. Many (sometimes even the majority) phd students in statistics are actually math majors who don't have much in the way of stats courses beyond introductory ones they took in college anyways (at least, that's what I've seen). It seems like you fit that description. I think your math gre score is also a plus. You should get into at least a couple places in your list though I'm not sure how competitive the UC schools you listed are.

I don't think you need to apply for masters programs since it looks like you have a strong profile already and you have taken so many courses/credits. I guess the only thing I would be concerned about is research experience, but otherwise you look pretty strong. I'm guessing you graduated outside of the U.S., which would make things a bit more competitive when applying to phd programs., but I think you should get in somewhere.

Yeah, I would just try to stick with it. I guess you can try to do well on the final to get a decent grade. Even if you just pass the course, you can always take some kind of extra analysis course (whether it be a grad or undergrad course) when you're in grad school to make up for any gaps in your knowledge (if there are any). I agree, though, that there isn't much information out there about admissions to grad schools in statistics. You can try looking at mathematicsgre.com which is for math phd/masters applicants but also has people talking about admissions to statistics programs. On their annual results page, you can usually see a couple of people who applied to schools in statistics and see what their profile/stats look like (and where they got in).

I don't think just having a drop on your transcript alone will hurt you too much. People drop courses all the time. However, I'm not sure what "advanced calculus" means? Does it mean like multivariable calculus/differential equations or an intro to real analysis? If it is the latter, then I think it would be important to stick with the course if you want to go to grad school in statistics since many schools require/recommend some form of real analysis. At the same time, the requirement might be less strict for masters programs compared to phd programs. At many masters programs, I believe some students have not taken real analysis and have the chance to take it for the first time through an undergrad course or some other means before taking the traditional stats grad classes.

What kind of math/statistics/computer science courses have you taken? I think it is better to get a LOR from a professor in math/statistics who can attest to your quantitative ability. Perhaps a masters program might be a better fit since PhD programs usually look for some research experience and strong academic records (especially in math classes).

Maybe you can set up an independent reading course with a math professor in a real analysis text? Most universities' math departments offer some form of independent reading courses and you could easily set one up with a professor and study the text yourself. Perhaps you can take the final exam of the regular real analysis class and still get credit for it? I think grad schools would understand if you had no choice but to take an independent course as long as you learned the material in the end.