Changing field from Economics to Statistics/Machine Learning

[manthew]

Hello Gradcafe. English is not my main language, so bare with me for a bit.

A little background about this post. I obtained my Bachelor in Economics last year and was enrolled to a PhD fast track of the same field in Germany. I'm thinking of dropping out of by PhD track and take specialisation courses with the Compsci and Stats department to complete my Master in the 5th Semester (2.5 years) rather than the usual 2-years and then apply for PhD in Stats/ML. The following are the reasons why I'm wishing to do this:

1.) In my first year in PhD track, I realised I want to do Statistics (specifically Bayesian nonparametrics) but because my department (medium sized compared to US) was going some massive disastrous exodus; there aren't many relevant professors left now. So in my specialisation phase, I'm confined to mostly microeconomics courses which I have no interest in.

2.) Three weeks ago, I attended a Statistics summer school but was exposed to Machine Learning at the same time and thought I could venture into this field and consolidate Econometrics and ML together. After reading Hal Varian's article and some googling on this.. I am convinced.

=====================================================

Now I know this would raised quite a some questions, so I will attempt to answer some of them here:

1.) Structure: So basically, PhD students must take at least 80% of their specialisation courses from the department (which I have no interest in). However, If I leave the track and join the Master track, I can take whatever I want from any departments (even from a neighbouring university) since I already fulfilled the basic requirement for Master (sans Thesis).

2.) Financial. As a fast track student, we don't get any stipend until year 3 anyway, so I'm fine in terms of financial stability.

3.) Mathematical background: I'm quite worried about this. Coming from Bachelor in Economics, we have relatively little preparation in math. In my first year of PhD I have exposed to Analysis, elementary Probability Theory and some mathematical methods from micro and macro, but I'm not as strong as a typical Math graduate. When I wrote to the relevant professors, they said the courses might a difficult for me but they are willing to give me a try since I did quite well in the math/stats subjects as mentioned.

=====================================================

 

Plan:

Summer

I don't have any jobs this summer so I intend to self-study the following subjects:

Real Analysis (using books by Terrence Tao and practice exercises in baby Rudin)
Intro Mathematics Statistics (book by Hogg et al/ Dudley)
Statistical Inference (book by Casella)
Theory of Probability (book by Durrett)

Machine Learning from Coursera

 

3rd Semester:

Intro PDE
Multivariate Analysis
Statistics II (Asymptotics of M-estimators, local asymptotic normal models; wont be able to take Stats I because the prof is doing it in German)
High-diemension Covariate Estimation (seminar)

 

4th Semester
Adv Analysis
Probability Theory  (I feel like I should first do Adv Analysis (which involves Lesbegue integration) before doing any Probability Theory but I won't be able to do it if I want to graduate in 5th sem)
Financial Time Series (seminar)
Stochastic Analysis OR Stochastic Differential Equation
(I also have TA responsibility in this semester for Economics Growth, but I heard it would be easy so I guess I'm ok)

 

Summer 2016

Internship + learn a new coding language.

5th Semester

Algorithm
Data Mining
Master Thesis
Apply for grad school (Dreaming about PhD Stats in Duke/ (Applied) Math in Bonn/ Study under Prof Teh in Oxford)

 

*the modules are all Master level courses.

=====================================================

Academic Results (for profile evaluation)

Undergraduate (from Russell groupin UK) BSc. Economics
Adv Mathematical Economics (A)

Applied Algebra (A)
Applied Calculus and PDE (A)
Applied ODE (B )
Mathematics for Engineers I & II (both A) 

*I left out other low level Math courses (eg year 1 math)

 

PhD Courses (from top Econ school in Germany)
Mathematics for Economists (Basically Analysis) - A+

Adv Econometric 1 (introductory probability/measure theory) - A-

Adv Econometrics 2 and 3 (both B ) 

*Passed my first year Quals

 

**Grade conversions might be a little off. 

 

=====================================================

Finally, my questions are:

1.) I would like to quit PhD Economics and obtain a Master of Economics (with 60% of my modules in Stats and Compsci(ML) courses). Do you think this is doable or I'm just crazy? (My colleagues say I am, but I really want to do statistics rather than economics, and I don't think I can last another year doing Microeconomics or Industrial Organisation when my heart is not there anymore)

 

2.) What would you think of my 'Plans'. Too unrealistic? 

 

3.) How would the Admission Committee see my profile? Favourable or unfavourable?

 

I would really appreciate your advice .Thanks

 

Edited July 10, 2015 by manthew