Suggestions on applied programs?

[Karoku_valentine]

Hi,

My situation is the following:

A) I finished my MS in Statistics in a theoretical program, where my classmates were PhD students in Stats. I passed my first qualifier with the PhD level, but I could not switch to the PhD on the second year because I did poorly in Probability (Measure Theory) B- (some people failed the course) and my Advanced Stat Theory. The program was really difficult to me, but I was able to get grades from B to A in the rest of the classes. Afterwards, I got a nice job and promotions in an international Financial company doing credit scoring and loss forecasting using SAS, R, Python and Matlab.

B ) I plan to get my company sponsorship in 3.5 years (they start the sponsorship in the year 4 after getting the H1B visa) to obtain the permanent residency/green card. So after that, I would like to return to a PhD, but I am uncertain whether I should pursue a PhD in Stats or in something more applied.

Pros for regular stats programs:

1. I already know the common materials for the first year and other second year classes (Casella Berger, Linear Algebra, Applied Bayesian, Applied Stats, GLM, Survival) and I have kept studying them since I use a lot of them in my current work. I think I can pass the Q1 of different schools since I had studied exams from several schools for my own Q1 (it didn't include measure theory, which was in the second year), and now I am much better at proofs and stats, but I feel very uncertain about being able to pull it off in a program that has Measure Theory in the first qualifier.

2. Since I have taken several stats classes, I have a better understanding of the program, so I can read more papers, take more classes, etc.

3. I would be competitive in a program with ties to finance, operational risk, extreme value distributions, Monte Carlo simulations, etc. since I have work experience on that. Though, I think working experience may be kind of irrelevant  for a PhD admission.

4. I plan to impress a colleague (not my boss, but a project manager who assigns me work cause my boss is too busy to assign me tasks), who's a PhD in Applied Math, so I could get a recommendation letter from him.

Note: I am not interested in academia and I understand my situation, so I would be thrilled to be in a 50+ program.

Cons regular stats programs:

1. Several programs include Probability-MT in their qualifiers, even the lower rank programs.

2. I have poor grades in the Probability Class and the Advanced Theory class. Therefore, several programs would automatically disqualify me.

3. My Analysis classes were not as rigorous (we did not use Rudin).

 

I have been thinking of more applied programs. Are there any statistics PhD programs that are more applied ranked 50+? Any suggestions? I don't think I could do biostats because I have no bio experience (or interest) to justify that.

I have also considered PhD's in Operations Research, where I think my working experience may make me more competitive and the most theoretical courses may have a lower impact. Though for OR I may need to take additional classes (linear programming, optimization, stochastic processes) before the program.

There's plenty of time (around 4 years to start applications), but I want to start preparing. My previous GRE was 168 in Math, so I would have to retake it in a few years.

Thanks!

 

Edited April 12, 2020 by Karoku_valentine

[statsday]

Do you mind me asking why you want a PhD? Not trying to talk you out of this just want to understand your goals more clearly. 

[Karoku_valentine]

11 hours ago, statsday said:

Do you mind me asking why you want a PhD? Not trying to talk you out of this just want to understand your goals more clearly. 

Yes, I have the following reasons:

1) Industry standards. I have noticed that there are several research positions in the financial sector that demand a PhD and to have knowledge of stats/optimization in credit scoring, fraud detection, insurance. Some positions explicitly request a PhD. They require to use techniques in journals or devise completely new methodologies, make sure algorithms converge, apply missing data techniques, optimization, Bayesian, and sampling techniques too. For example, my informal supervisor (my formal supervisor is a managing director) read papers from different journals and implemented a methodology to calculate losses that was validated by the federal reserve. It was not a ground-breaking methodology, but he had to read a lot of papers to have a better idea of what to do, and apply what they had in our context. 

2) Personal reasons. When my informal supervisor has given me papers, I notice gaps in my understanding of certain proofs, concepts and I confirmed there are so many things I could learn or study. Even if I keep studying and reading, I think I can learn a lot more in a real program since I would have feedback from professors and peers, be exposed to a more creative environment and specialize in a field.

3) Promotions. Some higher level positions do require a PhD, like the head of modeling, for example, where 4 out of 5 heads had a PhD in a Quant methodology. Or I can change fields, thrive and still have a similar salary. More positions will be opened that will require newer solutions and I want to have the skills. The PhD would give me access to positions that may not be in my reach right now.

 

[bayessays]

How was your PhD program ranked?  If you went to a highly ranked program and have good letters, it might be easier to drop to a lower ranked program.  There are plenty of stats PhDs even at top programs that don't make you take classes, yet alone tests, on measure theory.  Think places like Indiana U, George Mason, many others that are outside of top 50.

I think lower-ranked biostat programs could be a good option unless you really want to do something like financial research.  They don't require any biology background at all.

Then, there are some new PhD programs that are less traditional.  For instance, there are data science PhDs.  Alabama has an applied statistics PhD, and South Dakota State has a Computational Science and Statistics PhD (one of their graduates is actually a professor at Iowa State, a top 25 program now).

[Karoku_valentine]

1 hour ago, bayessays said:

How was your PhD program ranked?  If you went to a highly ranked program and have good letters, it might be easier to drop to a lower ranked program.  There are plenty of stats PhDs even at top programs that don't make you take classes, yet alone tests, on measure theory.  Think places like Indiana U, George Mason, many others that are outside of top 50.

I think lower-ranked biostat programs could be a good option unless you really want to do something like financial research.  They don't require any biology background at all.

Then, there are some new PhD programs that are less traditional.  For instance, there are data science PhDs.  Alabama has an applied statistics PhD, and South Dakota State has a Computational Science and Statistics PhD (one of their graduates is actually a professor at Iowa State, a top 25 program now).

Thank you for your answer. The school in which I was doing my MS in Stats, but all my classes were at the PhD level, was in the 20-30 rankings. I just feel that the B- is haunting me.

I just feel it would be "out of character" to have worked in the financial sector for around 8 years by the time I apply and say I am interested in Biostatistics (though, the classes I took that were shared with biostats were actually really good and interesting: GLM, Survival, Missing Data) when nothing in my profile screams biostats. I'll have to check if I can do something to make my profile look a little more bio.

I had seen the SDSU program and contacted one of their professors. He shared the syllabus for some classes, but did not share the qualifiers (some schools straight publish theirs online, I don't see why not) I think their qualifier still involved some Measure Theory, but I gotta pay more attention to that.

Please anyone else, let me know if you hear of any interesting applied programs. Thank you.

[bayessays]

It's totally fine if you are personally uninterested in biostatistics, but the programs will not care at all.  You have a strong math and stats background, and they will like that.  Most people going into biostatistics PhD programs have zero bio background.  Even people going into really bio-heavy subfields like statistical genetics often have no biology background.

 

[Karoku_valentine]

On 4/12/2020 at 4:09 PM, bayessays said:

It's totally fine if you are personally uninterested in biostatistics, but the programs will not care at all.  You have a strong math and stats background, and they will like that.  Most people going into biostatistics PhD programs have zero bio background.  Even people going into really bio-heavy subfields like statistical genetics often have no biology background.

 

Thank you. I think I'll try to take some courses online on Clinical Trials and Epidemiology at the U Penn State Global Campus.