Subscribe to DSC Newsletter

Are data science or stats curricula in US too specialized?

I earned my education in Belgium and never attended college in US, so it is hard for me to answer this question, even though all my professional experience and career took place in US. Yet sometimes people ask me: is your degree in stats, mathematics, or computer science? I have a hard time answering, because, in fact, it was in these three areas, each with equal weight. 

In US, my understanding is that you focus on one area in particular, when in college. This explains why sometimes, statisticians graduate with little programming experience, or being unfamiliar with machine learning techniques, data structures, or very applied stats. It also explains why recruiters think that data scientists are like unicorns. 

Of course over time true data scientists become masters, if not experts, in all these areas. But fresh, out-of-college data scientists lack this combined expertise. I am curious to see if my perception of the educational system (especially in US) is correct, and if yes, how it can be changed. Possibly, some of the issues stem from the fact that

  • university professors are highly specialized in a narrow field
  • faculty departments are very homogeneous: you rarely see a department or program that blends professors from computer science, mathematics, and stats -- they are assigned to separate departments

I think this is changing, albeit slowly. How can it be improved? For those who attended data camps or Coursera lectures to complement their knowledge and fill the gap, was it helpful? Are programs from top universities (say MIT) more balanced, like the education I received (up to my PhD) at University of Namur?

For related articles from the same author, click here or visit www.VincentGranville.com. Follow me on on LinkedIn, or visit my old web page here.

DSC Resources

Views: 324

Reply to This

Replies to This Discussion

I did not obtain my degree in the US so I won't be able to contribute to most of your question but I have spent a good amount of time expanding my curriculum with online resources like Coursera, edX, MIT opencourseware and a few others.

Context: I have a PhD in astrophysics from Université de Paris (XI), France. To get there I had significant education in stats, programming and signal processing on top of the core Math and Physics curriculum. I did spend 1.5 year of my PhD in Caltech (co-advisor of my PhD thesis was head of a lab there) and in general all my colleagues that were following courses there seemed more focused on the core curriculum. Programming was widely practiced and studied though but that might be specific to my field of study.

I personally feel that online courses have helped me fill gaps in my curriculum, especially on Machine Learning which I learnt as I needed it (worked with classification and regression during my time in astrophysics). Knowing a fair bit of the material when I enter a course also makes it easier to complete. I also use online courses to break into technologies new to me (that's how I started learning how to use Hadoop, Spark and Tensorflow for example).
I should also say that I have dropped out of a fair number of courses too due to low perceived return.

Hope this helps !...

Your education background is amazing Benjamin! Many data scientists don't have that kind of knowledge and expertise. I was too involved in a lot of coding during my college years and PhD (including C++) as well as signal processing. Not just little pieces of code, but full-fledged software with many thousands line of code, doing both low-level tasks (loading an image very fast in RAM) and high level (image segmentation.) That was back in the early nineties. Later on, I designed my own APIs to create taxonomies and for web crawling (e.g. to find copyright infringements, or to produce buy/sell trading signals.)

Benjamin Bertincourt said:

I did not obtain my degree in the US so I won't be able to contribute to most of your question but I have spent a good amount of time expanding my curriculum with online resources like Coursera, edX, MIT opencourseware and a few others.

Context: I have a PhD in astrophysics from Université de Paris (XI), France. To get there I had significant education in stats, programming and signal processing on top of the core Math and Physics curriculum. I did spend 1.5 year of my PhD in Caltech (co-advisor of my PhD thesis was head of a lab there) and in general all my colleagues that were following courses there seemed more focused on the core curriculum. Programming was widely practiced and studied though but that might be specific to my field of study.

I personally feel that online courses have helped me fill gaps in my curriculum, especially on Machine Learning which I learnt as I needed it (worked with classification and regression during my time in astrophysics). Knowing a fair bit of the material when I enter a course also makes it easier to complete. I also use online courses to break into technologies new to me (that's how I started learning how to use Hadoop, Spark and Tensorflow for example).
I should also say that I have dropped out of a fair number of courses too due to low perceived return.

Hope this helps !...

Reply to Discussion

RSS

Follow Us

Resources

© 2018   Data Science Central ®   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service