As someone who obtained their Master's in Physics, I have to agree with the article. The breadth and depth of my scientific training has made me a much more capable data scientist (and engineer) than most of my peers who went the CS path. I only wished that I could have double majored in biology. I would say that my study in physics has taught me how to better see the "bigger picture" and to try to account for all possible outcomes in a given scenario.
> As someone who obtained their Master's in Physics, I have to agree with the article. The breadth and depth of my scientific training has made me a much more capable data scientist (and engineer) than most of my peers who went the CS path.
Not knocking on you specifically, but I've seen this kind of overconfidence before in physicists, mechanical engineers, mathematicians etc.: "programming / statistics / data science is so easy, of course I know what I'm doing and will be up to speed right away." Their work is generally not bad, but not particularly great either. I do think the scientific training and rigorous education in these fields helps, but if it's not supplemented with a little humility it makes for poor colleagues.
It isn't overconfidence. The academic material these individuals study is directly and specifically related to the work of a data scientist. Statistical analysis is built in to any reasonably good physics curriculum starting with undergrad lab courses and proceeding through statistical mechanics, quantum physics and beyond.
I tend to agree. If I have a bunch of applicants for a programming job, given the choice between a similarly capable computer science graduate, or a natural sciences graduate (e.g. physics, chem, bio, oddly psychology - except I'm biased I have qualifications in the last two) then I'm likely to pick the non CS qualification applicant because of the breadth of experience, and the implicit recognition that the computer isn't the most important part of the job. Having said that I've worked with a few excellent CS grads, and for some tasks, good ones are really valuable.
That really depends on what you mean by "similarly capable."
If we're talking a psych grad who taught themselves the equivalent of a CS degree, then sure. That's usually not the case though. Non CS grads tend to never learn the boring parts.
As someone who worked professionally for almost a decade before getting my CS degree, what I learned during my degree has made a huge difference. There were so many gaps in my knowledge that I didn't know I had.
I've also hired plenty of CS grads, and non CS grads, and in my experience, the CS grads tend to outperform the rest all else being equal.
Are you sure that you're not seeing selection bias here? Non CS majors tend to be career changers (or they've had time to finish and degree and learn CS) and are therefore older on average.
How about this way of putting it: For many teams you don't need many people with a strong background in CS. Yes, strong CS grads can be useful, but often it's just as more valuable to have self-taught or similar (i.e. highly motivated) people with a breadth of experience outside the realm of having the computer as the primary focus in the task - because for many programming tasks, the computer isn't really the primary focus. A relatively small proportion of strong CS grads in a team can be really useful.
> do you think a physics degree or a statistics degree is better suited to doing data analysis, and why?
For finance, physics 100%. (The more experimental, the better.) When you're forced to map your theories to reality, and deal with the divergence, you develop an intuition for a certain set of problems. Those problems recur in commercial data science.
Reading through the thread, it looks like you believe that physicists can perform data analysis better than statisticians, because a physics program includes some stats courses.
