Learn at source

People sometimes ask me how they can learn the skills they need to work as a data scientist. My preference is to learn at source: learn each skill from the domain where it originated and where there are most clearly experts.

There are increasingly many data science course, which aim to teach the combination of maths, science and programming that is seen as key to data science. I am a bit skeptical of these online courses, if your goal is to be as good as possible at what you do.

Having identified the maths I wish to learn, I try to learn from a maths department. Learn programming from software engineers. Learn science by working with scientists and watching how they go about solving a problem. Learn each skill where there are experts with that skill; they will often have a clarity of understanding and mastery of the topic that goes beyond sources that have aggregated content together.

For this to work well, it helps for me to keep certain other things in mind:

  • Don’t be elitist

I prefer to learn from specialists of a skill, so am more likely to turn to a maths department lecture before a data science course, but this doesn’t mean favouring a hefty textbook over a succinct blog post. Clarity and depth of understanding are the speciality of experts, not reserved for academia or old-school teaching.

  • Teaching matters more

I might start by looking at maths courses, but I won’t be settle for dry and dour lectures just because they’re in font of a chalk board. I just happen to think that looking hard enough will usually turn up a good teacher (Gilbert String, Joe Blitzstein and Michael Nielsen are three teachers I appreciate a lot).

  • Experts don’t have to be narrow

One of my favourite resources to learn from is David MacKay’s “Information Theory, Inference and Learning Algorithms”, a book that navigates so many different topics I’ve no idea how to summarise it. He seems to have been someone who managed to truly understand and teach many things in combination.

Perhaps I just have a preference for understanding the fundamentals of a subject, and that’s more difficult when you move further from the source.