ar_rf

Most schools follow the guidelines set down by the April 15th Resolution, which you can read here. The relevant portion:

This is definitely what I would recommend. The best way to compensate for a low grade in an important math class is to excel in a harder math class. Especially because the math in your profile sounds a little borderline either way, and Real Analysis is a great boon to any profile and highly recommended if you're aiming at top programs.

I'm not very familiar with the masters program (you probably know more than I), but it seems like a pretty terminal program to me. Some masters programs are decent ways to buff up your PhD application, but Michigan's doesn't seem to be one of those. It might be possible to take some of the PhD classes and additional math, but it doesn't seem like it's standard and the terminal reputation of the program probably won't help too much in PhD applications. This is in contrast to some other masters programs that I've heard of, which are more connected to the associated PhD program. I don't think there will be a real difference for industry jobs between the MAS and getting a masters along the way at a PhD program. This probably depends more on the individual programs than anything, as most programs place very well into nearby industry. I think that 3) is probably an accurate statement. It's your decision, but I doubt that spending the extra year (and possible significant money) at a masters program would be worth it. If you have a solid funded offer, which is sounds like you do, I would recommend taking it.

Be careful if the professor that you want to work with at Florida is Michailidis. I'm guessing that he will be very busy, as he will be directing the new Informatics Institute and I've heard that he will still be remotely advising the students he had at Michigan (as the hire seemed to be sudden and unexpected, so he didn't unwind his students like people often do). You might want to confirm if he will be taking students if you can get in contact.

I guess I should also disclose that I have lived in Chicago for the last six years, so I have a pretty strong bias towards Chicago over Pittsburgh. Others might not have the same preferences that I do. I wasn't in this situation though, as I was not accepted at Chicago. I will be attending Michigan over CMU next year though. It was a hard choice and I loved both programs, but Michigan won out due to some great conversations with faculty and a location preference for Ann Arbor.

Ten or even five years ago, this probably would've been an easy decision to go to CMU. But Chicago has been hiring a lot of faculty in ML and related fields recently (including Lafferty, who is one of the top people to work with in the field), and I suspect that their near-future hires will continue that trend. Also, don't forget that TTIC is just down the street and offers great access to a variety of more compsci-minded ML researchers. The department is a bit late to the game, but they are making a big push into those areas and now is a good time to benefit from that. Chicago is also a much better place to live than Pittsburgh, and is one of the best cities to live in period. Going somewhere where you can be happy living for six years is important. CMU is obviously very strong at ML, but to me this would be an easy choice.

I very much got the impression that if you're a native english speaker then you are pretty much going to be able to get support through a TAship at Michigan. They have a lot of teaching that needs to be done and not nearly enough PhD students to do it.

I don't think you're aiming too high at all. I'm not particularly familiar with masters programs, but I imagine that you'd be very competitive at almost any program. One program you should definitely look at is Michigan, they have a strong ML group and most of their english-speaking masters students are funded as TAs.

There was a on this that I found very helpful, but I was hoping to get an updated (and conditional on my situation) perspective. So far, it looks like my top two choices are Michigan and CMU Stat (not CMU's joint stat/ml). My primary interest is machine and statistical learning, though I have a couple other related interests as well. I am not sure whether I would prefer academia or industry, but I would definitely like to keep my options open (i.e. aim for academia). CMU is generally ranked a couple spots above Michigan, and given my interests you'd think that it would be a clear choice. Indeed, when talking to various professors they usually seem to assume that I will be going to CMU. Would it be crazy to consider Michigan over CMU though? It seems to me that it might be just as good of a place to do ML and related work as CMU, especially if I am not in the Joint Stat/ML program. Even without access to ML and CS faculty, CMU Stats obviously has some great people working in that area. But to me it seems that Michigan has even more very good people working in that area. They also seem to be in a great place in the demographic cycle, with a ton of just pre- or just post-tenure faculty who are very productive and active. For me, the location also isn't a huge factor. I would love to hear anyone else's thoughts.

I've pretty much received all my decisions, and I suspect others have or will soon as well, so I might as well bump this. Undergrad Institution: Top 10 private Major(s): Statistics, Economics Minor(s): n/a GPA: 3.86 Type of Student: Male Domestic Asian GRE General Test: Q: 170 (98%) V: 167 (97%) W: 4.5 (80%) Programs Applying: PhD Statistics, Econometrics and Statistics Research Experience: Significant amount of econ research in school. Two years as a full-time RA at a Federal Reserve bank doing econ research. Letters of Recommendation: Good one from a well known stat professor, good one from a famous econ professor, great one from a relatively unknown econ researcher. Berkeley - Pending, presumed reject (as of 3/5) Washington - Waitlist (1/23) Chicago - Reject (3/5) Carnegie Mellon - Accept, no stat/ml (2/24) Michigan - Accept (2/9) Wisconsin - Accept (1/28) Columbia - Waitlist (2/26) Minnesota - Accept (1/21) Purdue - Accept (1/15) Ohio - Accept (12/30) UCLA - Accept (1/29) UIUC - Accept (2/16) Chicago-Booth - Accept (2/16)

To be honest, I would be very surprised if they didn't hold that time gap against you. A lot of things can change in 20 years, and with no recent evidence to the contrary adcoms are likely to assume that you skills and knowledge have declined at least a little in that timespan. Remember that most adcoms are very risk-adverse, which is understandable since a funded offer is often a ~$500k commitment on their part. If you do end up taking another year and can afford the cost, I would suggest looking into 1-year stats/math masters programs nearby, or at least try to take some classes as a non-degree student. Good grades in a couple difficult and/or graduate classes would probably help a lot. Excelling on the math GRE could also help assuage these concerns, as you already know.

Yes, very similar email to the one you got. I contacted them about it mid-last week. I also have the say, I think the tone of the original post is a little off-putting. You've gotten into a set of schools that most people would kill to have even a shot at, yet you're expressing frustration about getting waitlisted at a school that you admittedly wouldn't even consider. It's good advice for others to check their applications at CMU if they had the same issue, but some of the other stuff seems a little insensitive.

I was affected by the issue as well, and received the same message after contacting them. I was accepted to the Statistics program the other day though, so they definitely did give consideration to those affected.

I just checked and my ApplyYourself status still just says submitted, even though I got an offer letter several weeks ago.

My guess is that you're on an unofficial waitlist and that they're waiting for decisions to come back before putting out other offers. If true, that means it could be a while.

I think, pretty much regardless of industry, there's always going to be a significant local recruiting advantage. I don't think it'll be an issue to land an out-of-state job, but it's going to be significantly easier to do so with a more local company. This is probably even more true for internships that you might be able to do during your MS, which often lead to job offers once you graduate.

You can probably get an idea of their research profile by looking through their faculty pages. That description is mostly from what I've heard, I ruled them out early on from what I heard so I never really closely examined it.

If you're considering these schools, I would suggest just waiting for visit days to determine their research interests. It will be much more accurate and enlightening than anything you hear here. That said, these are some (possibly very inaccurate) general impressions I've gotten from my research during applications and second-hand reports. Despite being housed in a business school, UPenn Statistics is one of the most theoretical programs there is, almost to the point of absurdity. If you want to have any applied elements in your research, I would not necessarily recommend them. They are some of the best at what they do though, and if you are interested in theoretical work or math stat it's hard to beat them. Columbia has always had a strong finance emphasis, more than maybe any other school. They did hire David Blei awawy from Princeton recently, who is somewhat of an up-and-coming star in machine learning. I am unsure whether this signals a new initiative or direction in ML, but it is interesting. UChicago is a small but diverse department, with a couple people doing a bit of everything. They recently hired Lafferty away from CMU, who has been a fairly big name in Machine Learning for a while. (Blei was actually a post-doc under Lafferty at CMU, and I think many of the LDA extensions he did were with him.)

Same. Would love to hear about it.

Stanford is one of the only schools (the only?) that requires the math gre in the application (source). I've heard rumors that they've accepted applicants without a math gre before, but even if they are true I'm sure that it has to be an exceptional case.

Well, that wouldn't be the most surprising thing would it? I think a lot of traditional stats applicants either wouldn't know about or wouldn't necessarily want to go to a Econometrics and Statistics program housed in a business school. As far as I know it's a unique program. They probably mostly get econ-stat crossover types who might actually be more interested in the econometrics stuff. I don't know about their placements, but I also imagine there are worries that you'd have issues placing into a stat department. That said, I actually agree that they have a surprisingly good stat faculty with a lot of work being done outside of just econometrics. I also know that at least one of their more recently admitted students is pretty interested in Bayesian stuff, so maybe they're targeting different applicants? Or it's just one-year luck.

Small is an understatement; as far as I know they only take 1, maybe 2 people per year. I think that also means that it actually is quite competitive. I've heard good things about it though. I would think that it could be a bit lonely, but you have access to good faculty, there's a lot of faculty support, you have quite a bit of freedom to customize your curriculum, and because it's in the business school the funding is fairly generous. You also get at least some access to classes in the Statistics department. On the other hand, keep in mind that it is Econometrics and Statistics. So there's going to be more of an emphasis on the Econometrics part than a traditional Statistics program, which can be either a good or bad thing depending on your interests.

The thing is that if you aren't progressing statistical methodology in some way, then you usually won't be publishing in statistics journals, which pretty much disqualifies you from working in a statistics department. That applied of work gets published all the time, but it's usually in biology or psychology or economics journals. Of course, it's also a matter of definitions, it's possible that an Applied Statistics PhD is just on the applied side of statistics, but still very much in the field. I'm guessing that these things don't really have strict definitions. That being said, I don't think the title of your PhD will have any influence on your hiring prospects. Pretty much the only thing that matters at that point are the papers that you've published. If you can publish in top stat journals then you'll get hired. I think what's more important is looking closely at the program/department and the work that they are doing to make sure that they can give you the training and advising necessary to get those publications.

Econ research is different from stat research though. They do much more applied work, and in particular much more "purely" applied work with no (statistical) theory involved. I think that's rare in an academic stat department.

The April 15th resolution generally only applies to funded positions (and not all schools are signees, though most schools adhere to it). As far as I know, schools can set whatever deadlines they wish for all other positions.