Page 1 of 1

Stats PhD Evaluation- No idea where I stand

Posted: Sat Jan 23, 2016 12:55 pm
by choldawa
Can someone please weigh in and let me know if this is something I even have a shot at?
I'm a prospective PhD student in statistics hoping to get a better idea of my chances. I'm interested in Bayesian stats and its applications in machine learning.

I have no idea where I stand so any help would be greatly appreciated. Some preliminary data:

Undergrad Institution: US News 35-40
Major(s): BA Mathematics
Minor(s): Spanish
GPA: 3.9
Type of Student: Domestic White Male

GRE General Test:
Q: 164 (88%)
V: 162 (89%
W: 4.5 (72%)
GRE Subject Test in Mathematics:
Have not taken yet. Should I?


Programs Applying: UCLA (PhD stats), UCI (PhD Stats), USC (PhD Stats), Caltech (PhD applied and Comp. Math).
I'll be looking to apply this fall, and hoping to move to LA for personal reasons. Do I have a shot at these schools. Any suggestions of other programs I'd be competitive for if I don't end up in LA?


Research Experience: Just a Python-centric data science independent research course this semester. No publication.
Awards/Honors/Recognitions: Typical Dean's list etc.
Pertinent Activities or Jobs: Internships with BlueCross, Marriott, tech analyst. Served 2 year mission in Chile.
Letters of Recommendation: (3, personal, but none are particularly well known: algebraist, probabilist, machine learning)

My main concern is that I switched from a stats BS to a math BA and as such my coursework for a math major is pretty weak.
Coursework:
Calc 1-3 (A,A,A)
Linear Algebra and Diff Eqs (A)
Linear Algebra- Proof based (A-)
Honors Real Analysis (A-)
Probability (A)
Bayesian Stats (A)
Stats 1-2 (A,A)

THOUGHTS? Thank you so much!

Re: Stats PhD Evaluation- No idea where I stand

Posted: Sat Jan 23, 2016 2:43 pm
by arima
If you are interested in Bayesian statistics, then you should shoot for Duke U department of statistics. Kind of a Bayesian statistics Mecca. Duke states explicitly that they will not look at the GRE math subject test.

Re: Stats PhD Evaluation- No idea where I stand

Posted: Sat Jan 23, 2016 2:51 pm
by ghjk
arima wrote:If you are interested in Bayesian statistics, then you should shoot for Duke U department of statistics. Kind of a Bayesian statistics Mecca. Duke states explicitly that they will not look at the GRE math subject test.
Yeah, but I think we should advise him to hedge his bets. I'm not quite familiar enough with Stats department to give advice on which department is good, but he might want to try UCSB's Probability and Stats Phd program. I really think, through his coursework, and his subpar research experience, he doesn't stand a good chance at UCLA and no chance at Caltech. He might be waitlisted, then admitted at UCLA, if a random factor is highly positive for him (and same scenario would probably apply to USC, although I don't know how tough that department is at USC). From your description, Duke is probably too tough for him, but it never hurts to give it a try, so I would save money to apply for Duke rather than Caltech. The lack of graduate-level statistics course is a MAJOR weakness, as in his case, he unfortunately cannot make it up by saying "look, I make it up by doing good research on an advanced topics in Stats named bla bla bla, etc"

Re: Stats PhD Evaluation- No idea where I stand

Posted: Sat Jan 23, 2016 2:59 pm
by arima
ghjk wrote:
arima wrote:If you are interested in Bayesian statistics, then you should shoot for Duke U department of statistics. Kind of a Bayesian statistics Mecca. Duke states explicitly that they will not look at the GRE math subject test.
Yeah, but I think we should advise him to hedge his bets. I'm not quite familiar enough with Stats department to give advice on which department is good, but he might want to try UCSB's Probability and Stats Phd program. I really think, through his coursework, and his subpar research experience, he doesn't stand a good chance at UCLA and no chance at Caltech. He might be waitlisted, then admitted at UCLA, if a random factor is highly positive for him (and same scenario would probably apply to USC, although I don't know how tough that department is at USC). From your description, Duke is probably too tough for him, but it never hurts to give it a try, so I would save money to apply for Duke rather than Caltech. The lack of graduate-level statistics course is a MAJOR weakness, as in his case, he unfortunately cannot make it up by saying "look, I make it up by doing good research on an advanced topics in Stats named bla bla bla, etc"
I didn't catch the desire to be in California. He has a good chance at UCLA. UCLA's department of statistics is fairly new and not that strong, in the 30-40 ranking range. But, I wouldn't call UCLA a Bayesian focused school. A good option is UC-Riverside, there are some strong statisticians there who are Bayesian. Finally, Caltech doesn't make sense as a school for interests in statistics.

Re: Stats PhD Evaluation- No idea where I stand

Posted: Sat Jan 23, 2016 5:01 pm
by choldawa
Great, thank you. That's the first real answer I've gotten. I guess my follow up is if I'm competitive for masters programs as a transitional step. Is so, what schools in particular /what range of schools would I be competitive at?

Re: Stats PhD Evaluation- No idea where I stand

Posted: Sat Jan 23, 2016 6:17 pm
by MLHopeful
ghjk wrote:
arima wrote:If you are interested in Bayesian statistics, then you should shoot for Duke U department of statistics. Kind of a Bayesian statistics Mecca. Duke states explicitly that they will not look at the GRE math subject test.
Yeah, but I think we should advise him to hedge his bets. I'm not quite familiar enough with Stats department to give advice on which department is good, but he might want to try UCSB's Probability and Stats Phd program. I really think, through his coursework, and his subpar research experience, he doesn't stand a good chance at UCLA and no chance at Caltech. He might be waitlisted, then admitted at UCLA, if a random factor is highly positive for him (and same scenario would probably apply to USC, although I don't know how tough that department is at USC). From your description, Duke is probably too tough for him, but it never hurts to give it a try, so I would save money to apply for Duke rather than Caltech. The lack of graduate-level statistics course is a MAJOR weakness, as in his case, he unfortunately cannot make it up by saying "look, I make it up by doing good research on an advanced topics in Stats named bla bla bla, etc"
Don't listen to this guy, if you look at his past posts a few days ago he told someone else that getting into UCLA would be a crapshoot two days before they got in.

UCLA's stats department is nowhere near as strong as their math department, and you'd have a reasonable chance of getting in - you should certainly apply, and I think that's all that really matters. UCI and USC are weaker than UCLA and your chances would be correspondingly higher. Caltech is likely the most competitive of the four, and that's an applied math program which is pretty different from stats... Caltech doesn't have much stats going on, afaik. If you'd be willing to move to the east coast, Duke and Columbia may be worth applying to as 'reach' schools. Duke is certainly the most Bayesian school in the country (I've heard them described as 'cultish' by student that visited).

Re: Stats PhD Evaluation- No idea where I stand

Posted: Sat Jan 23, 2016 7:58 pm
by choldawa
MLHopeful wrote:
Don't listen to this guy, if you look at his past posts a few days ago he told someone else that getting into UCLA would be a crapshoot two days before they got in.

UCLA's stats department is nowhere near as strong as their math department, and you'd have a reasonable chance of getting in - you should certainly apply, and I think that's all that really matters. UCI and USC are weaker than UCLA and your chances would be correspondingly higher. Caltech is likely the most competitive of the four, and that's an applied math program which is pretty different from stats... Caltech doesn't have much stats going on, afaik. If you'd be willing to move to the east coast, Duke and Columbia may be worth applying to as 'reach' schools. Duke is certainly the most Bayesian school in the country (I've heard them described as 'cultish' by student that visited).
Thanks duly noted. I hope to apply and see what happens. Are there any masters I should consider given my goals and qualifications? Where do I stack up?

Re: Stats PhD Evaluation- No idea where I stand

Posted: Sat Jan 23, 2016 7:59 pm
by arima
MLHopeful wrote:
ghjk wrote:
arima wrote:If you are interested in Bayesian statistics, then you should shoot for Duke U department of statistics. Kind of a Bayesian statistics Mecca. Duke states explicitly that they will not look at the GRE math subject test.
Yeah, but I think we should advise him to hedge his bets. I'm not quite familiar enough with Stats department to give advice on which department is good, but he might want to try UCSB's Probability and Stats Phd program. I really think, through his coursework, and his subpar research experience, he doesn't stand a good chance at UCLA and no chance at Caltech. He might be waitlisted, then admitted at UCLA, if a random factor is highly positive for him (and same scenario would probably apply to USC, although I don't know how tough that department is at USC). From your description, Duke is probably too tough for him, but it never hurts to give it a try, so I would save money to apply for Duke rather than Caltech. The lack of graduate-level statistics course is a MAJOR weakness, as in his case, he unfortunately cannot make it up by saying "look, I make it up by doing good research on an advanced topics in Stats named bla bla bla, etc"
Don't listen to this guy, if you look at his past posts a few days ago he told someone else that getting into UCLA would be a crapshoot two days before they got in.

UCLA's stats department is nowhere near as strong as their math department, and you'd have a reasonable chance of getting in - you should certainly apply, and I think that's all that really matters. UCI and USC are weaker than UCLA and your chances would be correspondingly higher. Caltech is likely the most competitive of the four, and that's an applied math program which is pretty different from stats... Caltech doesn't have much stats going on, afaik. If you'd be willing to move to the east coast, Duke and Columbia may be worth applying to as 'reach' schools. Duke is certainly the most Bayesian school in the country (I've heard them described as 'cultish' by student that visited).
Hi @MLHopeful. Since you quoted me and @ghjk, he/she might think the "guy" you are talking about is me (@arima). The "guy" you are talking about is @ghjk.

Anyway, your message is right on mark and is consistent with what I said. I totally agree with you. Many people are unaware that UCLA's stat department became a stand alone department sometime in the 1990's. As such, it has not built a reputation and doesn't have a strong record of placement. The math department is a whole another story. As you and I said, Caltech isn't a choice for statistics. I do believe UC Riverside should be considered (one of the Bayesian legends -- Barry Arnold -- is there).

As far as master's degrees, Berkeley has an MA in statistics. That MA can be used as a stepping stone to better departments down the road.

Re: Stats PhD Evaluation- No idea where I stand

Posted: Sat Jan 23, 2016 8:08 pm
by arima
ghjk wrote:
arima wrote:If you are interested in Bayesian statistics, then you should shoot for Duke U department of statistics. Kind of a Bayesian statistics Mecca. Duke states explicitly that they will not look at the GRE math subject test.
The lack of graduate-level statistics course is a MAJOR weakness, as in his case, he unfortunately cannot make it up by saying "look, I make it up by doing good research on an advanced topics in Stats named bla bla bla, etc"
@ghjk, this comment shows that you are unaware what statistics departments are looking for. Statistics departments don't care too much about coursework taken as an UG in statistics (whether UG level or Graduate level). Statistics departments would rather see math courses first and computer science courses second.

Re: Stats PhD Evaluation- No idea where I stand

Posted: Sat Jan 23, 2016 8:13 pm
by ghjk
MLHopeful wrote:
ghjk wrote:
arima wrote:If you are interested in Bayesian statistics, then you should shoot for Duke U department of statistics. Kind of a Bayesian statistics Mecca. Duke states explicitly that they will not look at the GRE math subject test.
Yeah, but I think we should advise him to hedge his bets. I'm not quite familiar enough with Stats department to give advice on which department is good, but he might want to try UCSB's Probability and Stats Phd program. I really think, through his coursework, and his subpar research experience, he doesn't stand a good chance at UCLA and no chance at Caltech. He might be waitlisted, then admitted at UCLA, if a random factor is highly positive for him (and same scenario would probably apply to USC, although I don't know how tough that department is at USC). From your description, Duke is probably too tough for him, but it never hurts to give it a try, so I would save money to apply for Duke rather than Caltech. The lack of graduate-level statistics course is a MAJOR weakness, as in his case, he unfortunately cannot make it up by saying "look, I make it up by doing good research on an advanced topics in Stats named bla bla bla, etc"
Don't listen to this guy, if you look at his past posts a few days ago he told someone else that getting into UCLA would be a crapshoot two days before they got in.

UCLA's stats department is nowhere near as strong as their math department, and you'd have a reasonable chance of getting in - you should certainly apply, and I think that's all that really matters. UCI and USC are weaker than UCLA and your chances would be correspondingly higher. Caltech is likely the most competitive of the four, and that's an applied math program which is pretty different from stats... Caltech doesn't have much stats going on, afaik. If you'd be willing to move to the east coast, Duke and Columbia may be worth applying to as 'reach' schools. Duke is certainly the most Bayesian school in the country (I've heard them described as 'cultish' by student that visited).
Are you serious with your accusations? I did say that that guy had a chance at UCLA:P And you should not judge my evaluation skill based on 1 (yeah, 1) single result alone, while I got the correct predictions on other profiles (for example, that of Fw1812) Also, prediction means there would be "true or false," so you shouldn't expect me to be correct 100% of the time. Why don't you post your profile and let me judge it?!

Re: Stats PhD Evaluation- No idea where I stand

Posted: Sat Jan 23, 2016 8:14 pm
by ghjk
arima wrote:
ghjk wrote:
arima wrote:If you are interested in Bayesian statistics, then you should shoot for Duke U department of statistics. Kind of a Bayesian statistics Mecca. Duke states explicitly that they will not look at the GRE math subject test.
The lack of graduate-level statistics course is a MAJOR weakness, as in his case, he unfortunately cannot make it up by saying "look, I make it up by doing good research on an advanced topics in Stats named bla bla bla, etc"
@ghjk, this comment shows that you are unaware what statistics departments are looking for. Statistics departments don't care too much about coursework taken as an UG in statistics (whether UG level or Graduate level). Statistics departments would rather see math courses first and computer science courses second.
That's good to know. But his math courses are not strong as well, although it suffices the requirement of statistics department.

Re: Stats PhD Evaluation- No idea where I stand

Posted: Sat Jan 23, 2016 9:46 pm
by MLHopeful
arima wrote:
Hi @MLHopeful. Since you quoted me and @ghjk, he/she might think the "guy" you are talking about is me (@arima). The "guy" you are talking about is @ghjk.

Anyway, your message is right on mark and is consistent with what I said. I totally agree with you. Many people are unaware that UCLA's stat department became a stand alone department sometime in the 1990's. As such, it has not built a reputation and doesn't have a strong record of placement. The math department is a whole another story. As you and I said, Caltech isn't a choice for statistics. I do believe UC Riverside should be considered (one of the Bayesian legends -- Barry Arnold -- is there).

As far as master's degrees, Berkeley has an MA in statistics. That MA can be used as a stepping stone to better departments down the road.
Woops, right you are - I wrote that on my phone and was certainly referring to ghjk. I see he replied, but I'm not going to engage - just read his comments for yourself and come to your own conclusion (I was referencing this thread: http://www.mathematicsgre.com/viewtopic.php?f=1&t=3609, where he did in fact use the word 'crapshot').

As for masters programs, I think it's safe to say that masters programs as far less competitive than PhD programs and if you're looking to maximize name branding than that is undoubtedly the best path to take. Since they're mostly course-based, your lack of research won't hold you back, but unless you manage to get involved in some research I'm not sure how useful it'd be for continuing on to a PhD (plus they're expensive). If you have any interest in applications to biology, I think biostats programs tend to be less competitive; UC Berkeley is probably the closest reputable program to LA.

As a rough ballpark of department caliber, I'd recommend looking at the US news rankings. Although they're certainly not perfect (I'd put UC Berkeley stats at #1 :) ), they're good as a rule of thumb, taken with a grain of salt.

Of course, it's hard for me to say anything more concrete beyond that because I don't really know you, your LOR's or the application system terribly well (despite having successfully navigated it last year). In particular, I never considered masters programs, biostats, or PhD programs outside the top 10. The people best suited for these conversations are your letter writers.

Re: Stats PhD Evaluation- No idea where I stand

Posted: Sat Jan 23, 2016 10:23 pm
by arima
MLHopeful wrote:
arima wrote:
...I think biostats programs tend to be less competitive; UC Berkeley is probably the closest reputable program to LA.
You bring a good point about biostats programs. But, actually UCLA's biostatistics department is quite respectable and much more highly regarded than the cross-campus department of statistics department.

What is annoying about the US News ranking of statistics departments is that it mixes standard statistics departments with biostatistics departments. That is why one sees two Harvards (at #3 and #7) and two Washingtons (at #3 and #7) and Hopkins is listed #5 but it isn't statistics. The #3 for Harvard and U of Washington is not statistics but again biostatistics.

Why in the heck doesn't US News properly separate. Based on its rankings after separation:

Statistics:

1) Stanford
2) Berkeley
3) Chicago
4) Harvard
5) U Washington
6) Carnegie
and so on.

Biostatistics:

1) Harvard
1) U Washington
3) Hopkins
and so on.

I agree with you that Berkeley can legitimately be considered #1. When it comes to theoretical statistics, I would order them as Berkeley, Chicago, and then Stanford. Harvard is much too overrated. Hardly any academic placements of their graduates and a very applied oriented department.