Advice on PhD in Stats & Duke's MSEM

[crlee]

I plan to apply for PhD in statistics and some masters, including Duke's Master's in Statistical & Economic Modeling, for Fall 2016.

 

I become interested in statistics after I began to write for a tech blog and understood the application for stats/ML. Watching some of the conferences in open data, computation & journalism also reinforced my interest in learning more statistics and programming.

 

I want to get advice on how to prepare myself this year, whether to take more math classes, do some research project, or work as RA or industry.

 

School -  Top 25 National, top 5 Public University

Major - BA in International Relations and Economics, Minor in Statistics (Graduated June 2014)

GRE - Q 168, V 166, AWA 4.5

GPA - 3.54/4.00

Demo - Asian, Domestic, Male

Math/Stats Courses

1. Multivariable 1 - A
2. Multivariable 2 - B+
3. Linear Algebra - A
4. Probabilities - C
5. Mathematical Stats - A
6. Intro to R programming - A-
7. Regression Analysis - B-
8. Experiment Design - B-
9. Computation&Optimization - B+
10. Data Mining - B

I took the practice test for GRE Math subject test, scored around 50 percentile.

 

Yea.. my grades are not that great. 

 

Since I graduated this June, I've put more time and efforts into learning math and stats programming.

 

I am self-learning Real Analysis with Harvey Mudd's youtube videos. I find it challenging, but do enjoy it - which made me think I should do stats despite mediocre grades.

I've completed half of the JHU Data Science Specialization via coursera. I'm doing distance learning from Harvard Data Science course. I'm also sitting in a numerical analysis class at a public university, not my alma mater, near my home.

 

Research Experience - Not sure if these count ?

1. Senior thesis on economics - I used some survey analysis, and collected&analyzed firm-level dataset. It's not very quantitative oriented.
2. Research assistant for an education/statistics project - I was mainly responsible for data cleaning and producing statistical graphs (STATA and R). 

Letter of Rec

1. My senior thesis advisor would likely write a strong one, but my thesis is not heavily econometric.
2. Stats/Education Professor I worked for as a research assistant - decent letter, she was quite satisfied with the work I've done. But it was mostly following instruction - little independent work there.
3. Political Economy professor - I got an A+ and went to his office hours quite a few times. Again, not math or stats.
4. Stats Professor - I often seek advice from one professor, but didn't really do well in his class. I wonder if it's a good idea to ask for a letter from him.

Research Interests

Methodologies I'm interested in include bayesian, spatial, visualization, machine learning

 

Application areas

highly interested in - open data, data journalism, economic modeling, behavioral economics, social network analysis, business operation

mildly interested in - genomics, medical diagnostic (my brother works in this area, and I got exposed to this a lot at dinner table talks), financial modeling, natural language processing

 

Programming Experiences

1. Stats program - R and STATA
2. I'm learning python through Harvarvd CS 109 Data Science course

Questions

I have a year between now and Fall 2015. How should I spend my time learning and working on to improve my application and validate my interests in grad school?

 

My current plan is to complete JHU and Harvard's Data Science courses by Dec, and to take two math classes, real analysis and DE at a public university next Jan.

I think those two math classes would help me get good letter of recommendation and help me prepare mathematically.

1. What are other classes I should consider taking besides RA?

 

2. Would it help for grad school application to do side projects and put it on github? Is it real research experience?

3. And, what's the best way to learn about subfields, going to R or python meetup, watching conference talks?

 

Thank you!

Edited November 7, 2014 by crlee

[sild]

If you could get research experience from someone who could write you a strong letter of recommendation, that would be ideal.  One point to consider is that the Duke STAT department is highly specialized in Bayesian inference.  If you were applying to the doctoral program and did not mention Bayesian inference in your statement of purpose, you'd have a much lower chance of being accepted. 

 

You can always ask your stats prof if they feel comfortable recommending you for grad school right now, or whether you should get more experience/take classes.  If you take their advice and show them you're working hard, that could turn into a really nice letter.

[crlee]

Thank y

 

If you could get research experience from someone who could write you a strong letter of recommendation, that would be ideal.  One point to consider is that the Duke STAT department is highly specialized in Bayesian inference.  If you were applying to the doctoral program and did not mention Bayesian inference in your statement of purpose, you'd have a much lower chance of being accepted. 

 

You can always ask your stats prof if they feel comfortable recommending you for grad school right now, or whether you should get more experience/take classes.  If you take their advice and show them you're working hard, that could turn into a really nice letter.

 

Yea, thank you for the tip on the need to mention Bayesian for Duke's application. 

 

Gonna work on getting some research projects in stats/biostats area!