Transition into data science/machine learning from non-CS bachelor's degree

[physics2ds]

I'm interested in pursuing a PhD&career in either data science or machine learning in the future, but don't have an undergrad degree in CS, data science, machine learning, etc. and discovered my interest late so I've only taken 1 intro programming course and I'm assuming they have pretty rigorous pre-reqs

 

I majored in Physics at Cornell (cum.~3.6, 3.3 in major) and have had a couple years of research experience in multiple settings (working on first publication). I'd be willing to take a masters program in between, but wondered what my best option would be to make the transition given my relatively limited background.

[ErdosJr]

Getting an industry job as a data scientist wouldn't be too difficult given your background honestly. The skillset required is not by any means "rigorous". 

If your looking into graduate school, look into applying into MS Stats programs that have a heavy machine learning focus. MS programs generally don't have any hard pre-requesites. I did a double major in Math & Physics with 0 ML/Stats/CS classes, and it didn't affect my application much. Just be able to clearly demonstrate a purpose and interest in the field when you apply. 

[TakeruK]

Tons of people I know in all sorts of quantitative science fields, including, physics, astronomy, planetary science, environmental science, etc. have finished their PhD in these fields and gone on to data science positions in North America. Many more make this transfer while a postdoc and I know a couple of people who have tenure-track positions and decided to switch. In addition, a PhD is often not required (as @ErdosJr said) as I know a few people who were hired by data science companies while in the middle of their PhD and decided to leave the program for the job.

Data science is less about programming than it is about knowing how to think about data and working with data. The things that all of the above people had in common was that they took time on their own to learn key skills for data science. In particular, they all took some sort of online course on machine learning and had spent time creating some personal project using these skills that they can bring to the interviews. Some students took a more formal approach and officially added a CS minor to their PhD degree. 

I think that if you are dead set on a career as a data scientist, I don't think you need any more schooling and paying for a Masters may not be worth it. There are tons of guides online about transitioning from an academic background to a data scientist so I'd suggest following those. Spend time self-teaching and creating your own projects/solutions. On the other hand, if you are interested in both academia and data science and haven't decided yet, then I'd say you should follow the academic route for now (e.g. apply for funded PhD programs) and develop your data science skills on the side. Then, you can either leave your program or switch after graduation, depending on how things go and what your priorities become.

[dilemmas]

On 2/24/2018 at 5:55 AM, physics2ds said:

I'm interested in pursuing a PhD&career in either data science or machine learning in the future, but don't have an undergrad degree in CS, data science, machine learning, etc. and discovered my interest late so I've only taken 1 intro programming course and I'm assuming they have pretty rigorous pre-reqs

 

I majored in Physics at Cornell (cum.~3.6, 3.3 in major) and have had a couple years of research experience in multiple settings (working on first publication). I'd be willing to take a masters program in between, but wondered what my best option would be to make the transition given my relatively limited background.

I majored in Law, then did a masters again in Law, and after that did a PhD in Law, now I am starting a data science masters

I think you have much more background than me. So you might have a better chance of getting accepted to a PhD program. But I think you need to work on programming more.