CMU or UChicago (Stats PhD)?

[matcio]

Dear all,

I was accepted both at the University of Chicago and at Carnegie Mellon University in their PhD Program in Statistics.

These two programs are actually my favorites in the pool of those to which I applied.

I participated to the visit days and had a great time at both places; now it is time for me to try and make a final decision between and I am really having a hard time in choosing between the two universities: I enjoyed very much CMU's balance between theoretical and applied stats and also their focus on machine learning, nonparametrics and bayesian (these are all topics that very much appeal to me); also I had the feeling of a very nice department where there is a lot of interaction between graduate students and professors. On the other hand Chicago impressed me with a program that gives students a strong theoretical background during the first year with sequences in stochastic processes, probability theory and mathematical statistics (plus some applied statistics as well).

Both schools are excellent and students have excellent placements.

My impression, after the conversations I had with professors and students, is that UChicago is somehow slightly more theoretically-oriented while CMU is more concerned with applications, but this impression might be not completely correct and possibly biased by the particular set of professors with whom I had the chance to talk.

No doubt that there's a big difference between the campuses and the cities are completely different (Chicago is probably more attractive!), but those are aspects which I don't want to weight too much when making my decision.

Advice, comments or suggestions are most welcome!

Thanks

[ANDS!]

I would pick Chicago. The department is quite large so I would say you would be hard pressed to NOT find an advisor for your particular research interests once you reach that point. From a quick peek-a-boo to UChi's website, there are several faculty interested in Machine Learning, so I don't think you would be left out of the cold. Plus, cmon. . .Chicago!

Realistically, your dilemma is "Am I going to be boned come research time. . ."; I would think no.

I assume you have until April 15th - shoot an email to the chair and just do a bit of digging: "Thank you for the interview session, however I was hoping you could point me in the direction of any faculty who have X-interest and the current work they are doing with this topic."

[matcio]

Thank you very much for your reply ANDS!.

It's just that after the visit at CMU I was like "hey, this is the place for me!". Then I went to UChicago which somehow didn't click with me as CMU did, although I had a great time there as well and I felt like the overall quality of the university is higher.

I think my dilemma is more in the following terms: "do I prefer joining a department that (almost) surely matches my interests and in which it seems I would feel more comfortable or joining a more prestigious (perhaps?) and larger department where my first impression was not quite as exciting?".

Of course I see that I shouldn't make this kind of choice on the basis of first impressions, but I guess it isn't easy to totally ignore them as at least they play their role at an unconscious level.

Another plus for UChicago is that I feel I would benefit a lot from receiving a little more of tough theoretical training and they definitely provide that during the first 1-2 years.

Buridan's ass died essentially because he had not coins or dice... should I randomize my choice?

[ANDS!]

I think what are you are describing in terms of Chicago is emotional fit. Academic fit is all well and good, but overall fit encompasses much more, and if you didn't "feel" it during your preview week, which I assume included pow-wowing or being introduced to a good number of the faculty, then that is certainly a red flag that should go up.

Personally, I would still pick Chicago. Obviously CMU is also crazy well regarded, but Chicagos stock when it comes to their Stats program can not be sneezed at - it is simply one of the places you go if you are wanting a stats PhD. Plus, again, it's Chicago - I don't know if your long term goals are academic, but I should think you'd have a much easier time finding post grad (hell even near final grad) employment/research opportunities in Chitown than in Pittsburgh.

[qqyyzz]

Congrats on such nice placements. I think if you're even remotely interested in machine learning or data mining go with CMU by far.

[matcio]

Thank you very much to both for your kind and useful replies.

There are lots of pros for both schools, I'm finding it quite hard to decide.

I went again through the faculty webpages of both schools and it looks like there are really no problems of finding an advisor working in the fields that I am interested in the most.

One of my concerns is that I would feel more confident if I could take at least one more year of theoretical courses in probability, measure theory, stochastic processes and mathematical statistics; with regards to this aspect UChicago seems to provide a stronger training.

@ANDS!: yes, my long term goals are academics indeed.

[pyh474]

Congratulations! If you are particularly interested in machine learning, nonparametric, or bayesian, you should definitely go with CMU.

It's heard that it is not safe in the University of Chicago as the school is surrounded by very scary neighborhoods. Though this may not be something you would weigh too much when making a decision.

Both departments are fantastic, you would be happy at either school. Congratulations!

[Almagest]

I think you really should go to CMU if you are interested in machine learning and related topics. Their advantage over Chicago in this area is tremendous. It is indeed my top choice, but sadly I got rejected. Also, I talked to one of my professors, he told me that Chicago Stat department is not such a large department. There are many affiliated faculty members in that department. The number of core faculty members is quite small. Besides, I heard from some current statistics graduate students that the atmosphere in Chicago Stat is not so good. Students there are under a lot of pressure and many of them are quite depressed.

Anyway, congratulations on your offers, and hope you make a wise selection!

Edited March 27, 2012 by Almagest

[wine in coffee cups]

Fourthed(?) CMU.

[danthechub]

While this sounds like a tough decision, you can't really go wrong. I would say that the difference in prestige is pretty negligible, and both programs have a strong mix of theory and application (even though Chicago's core requirement is more theoretical). If you think Chicago is a better place to live but CMU is a better research fit, I would say CMU for sure. One of the bigger differences between the two is that CMU is significantly stronger in those areas you mention like machine learning. Good luck with your decision and you will be doing well either way!

[matcio]

Thank you very much guys for your helpful comments and for the very nice words!

I confess I am leaning a bit towards CMU, although UChicago is soooo appealing. To take this final choice is such a difficult task!

[gtg387y]

I'm curious as to why you applied to 22 schools, including such heavy-hitters as Chicago, Harvard, and Washington, and did not apply to Stanford or Berkeley.

That seems kind of insane to me.

[matcio]

Hi gtg387y,

I applied to a lot of schools to be reasonably sure that I could start a phd in the US.

Stanford required GRE subject (which I haven't take); both schools were simply too much for me however.

[gtg387y]

Hi gtg387y,

I applied to a lot of schools to be reasonably sure that I could start a phd in the US.

Stanford required GRE subject (which I haven't take); both schools were simply too much for me however.

Ok, now I get not applying to Stanford, but I think you sold yourself short by not applying to Berkeley. Obviously you were an attractive candidate! In any case, CMU and Chicago are both excellent schools. I would be ecstatic if I got into CMU (machine learning and data mining is my area of interest).