Profile Evaluation - Stats PhD 2022/2023

[raptorsfan]

Institution: Ivy (don't want to say which but not HYP)

Degrees: BS Computer Science/MA Mathematics (I'm currently on pace to finish in 3 or 3.5 years -- should I do this?)

GPA: 3.93 overall/4.00 major

Type of student: Chinese/Central Asian international student

Languages: English, French, Chinese, one other language I don't want to say (all native)

Awards: math olympiad stuff from high school

GRE: Haven't taken it yet

Programs applying: Statistics/ML

Research experience:

• CS research in high school (basically just got coffee)
• Deep learning research in CS
• Currently starting stats/ML research

Relevant coursework:

• Math: real analysis (graduate), abstract algebra (graduate), other undergrad-level courses
• Statistics: mathematical statistics, PhD ML seminar, other undergrad-level courses
• CS: ML, algorithms, etc.

Pertinent activities or jobs:

• Data science internship
• TA for CS courses

Applying to where:

• Top 20 stats programs

[Ryuk]

People with more experience will probably chime in soon, but how in-depth is your research? What are your letters of rec going to look like? 

If you have a decent application but all three letters are from superstars promising that you are a future COPSS winner, it would make a significant difference. Also, "deep learning research in CS" could mean barely more than a directed study or it could be several published papers. 

Information like this will probably allow the more knowledgeable people to make better predictions.

Edited March 8, 2021 by Ryuk

[dobzhansky]

Research/Internships:

Make sure you have something to show for your research; if necessary, take on smaller, more manageable projects that can be feasibly wrapped up by the end of this year and published. Apply broadly for internships and REUs; target universities or national labs to stand out from the "I interned at google analytics" crowd. This will also help net you better recs (a rec from a well-known prof or researcher is better than a rec from your team leader at some FAANG internship).

Awards:

Work on narrowing down your research interests and apply for more fellowships. You should specifically target the Goldwater and Astronaut (if offered at your institution). Fulbright or Knight-Hennessy also good. These are not only good exercises in writing about your research interests, but will hopefully help you get a better sense of the type of research you want to pursue. 

Otherwise, your academics seem pretty stellar and you should be competitive at basically every top 10 stats program. The differentiating factor will likely be how coherently and engagingly you can talk about your research interests and projects. Start following big-wig academics now and get a sense of their research so you can hold an intellectual conversation with them if they happen to interview you.

[raptorsfan]

@Ryuk Thank you for the advice! I haven't specifically asked any professors, but there are four who know me very well (two are well-known academics with Wikipedia pages and two are in their early 30s). In terms of research, I hope to publish a paper in deep learning/CS by the end of the summer (it's more on the applied side unfortunately). Also, I am working with two professors on statistics research. The work is highly theoretical, and I think I can publish results within the next 18 months.

@dobzhanskyThank you for your feedback! In terms of fellowships, other than Knight-Hennessy, my understanding is that they are only available for US citizens/permanent residents (I'm international). Will it be held against me if I do not have any prestigious fellowships?

Edited March 8, 2021 by raptorsfan

[Ryuk]

@raptorsfan Sounds like you're in pretty good shape! Again, I want to emphasize that I am just an applicant and in general you should place much more weight on the opinions of faculty or current PhD students. I know that some very young people on this forum make bold statements with even more confidence than bayessays or Stat Assistant Professor, who actually know what they are talking about.

All I can say is that I have similar letters, GPA, and research as you, and I was admitted to 8 of the top 10. I had no prestigious fellowships, but I did have awards for my performance in math classes. Virtually nobody can count on getting into the very top programs, but you should definitely apply to them.

[trynagetby]

If you got A's in your graduate real analysis class + you have a good relationship with 2 of your letter writers you're in good shape. I'm fairly confident that just having a good relationship with your profs so they write LOR is more important than publishing.  Given my previous setence you have a strictly stronger profile than me and I was got into Umich/Duke/Wisconsin/Uwashington.  Unfortunately I didn't apply to Harvard/Berkley (regret it now lol). So I think your floor should be the schools mentioned above and apply to all the schools above.

Edited March 8, 2021 by trynagetby

[kingduck]

9 minutes ago, trynagetby said:

I'm fairly confident that just having a good relationship with your profs so they write LOR

I second this. My GPA and test scores were on the lower end of the spectrum of gradcafe posters, and I still got into a good programs. I highly suspect this was due to 1) my LOR's strongly believed I was capable of a PhD, and 2) they are decently well-known in their respective fields. 

Edited March 8, 2021 by kingduck

[BL4CKxP3NGU1N]

Adding to the posts above, my GPA is high, but my GRE Q is fairly low, and I also have a pretty limited math background (currently taking real analysis). However, I believe my LoR's were quite strong, even though none of my LoR writers are very well-known. On paper, your profile is far superior to mine, and so far I've been accepted to a couple top 30 schools (CSU and OSU, based on US News rankings). I believe you will certainly be a competitive applicant for top 20 programs.

Edited March 8, 2021 by BL4CKxP3NGU1N

[Stat Phd]

Pretty sure you will get to many of the top 20 schools.

[dobzhansky]

4 hours ago, raptorsfan said:

@Ryuk Thank you for the advice! I haven't specifically asked any professors, but there are four who know me very well (two are well-known academics with Wikipedia pages and two are in their early 30s). In terms of research, I hope to publish a paper in deep learning/CS by the end of the summer (it's more on the applied side unfortunately). Also, I am working with two professors on statistics research. The work is highly theoretical, and I think I can publish results within the next 18 months.

@dobzhanskyThank you for your feedback! In terms of fellowships, other than Knight-Hennessy, my understanding is that they are only available for US citizens/permanent residents (I'm international). Will it be held against me if I do not have any prestigious fellowships?

As an international student, it won't really be held against you. But if you can, visit your school's fellowship office and see what might be on offer for international students. Anything that indicates your dedication to research counts. Like others have said, you're competitive for top 10 schools as it is; this would just be an extra leg up. 

[raptorsfan]

@Ryuk Thank you so much! Also congratulations!! Getting into those schools is super impressive. My school does not have awards for classes, but I have participated in my country's national math olympiad in high school, and I've been practicing the Putnam. Can I make up for not having math awards for classes with math contest stuff?

@trynagetby and @kingduck and @BL4CKxP3NGU1N and @Stat Phd Thank you so much for the input! I will make sure to have good relationships with my profs.

@dobzhansky That makes a lot of sense. I will definitely look into it.

Thank you so much to all of you!

[Ryuk]

@raptorsfan I'm sure you'll be fine without them. I think everyone agrees that your research and your letters of rec are the most important parts of your application. I highly doubt that my awards made a huge difference. You also go to a much better university than I did and that probably counts for something.