Learning Data Science: theories or examples?

What is a great learning process in data science? Traditionally it starts with theories behind an algorithm, then the mechanics of it and finally by exercising with one or two examples. Unfortunately this approach rapidly kills the excitement of learners, who loose interest in that algorithm real fast. They never get to the exitment of solving problems. Question is – is this the most efficient way to learn something in data science? Probably not.

I believe the most efficient way of learning data science is by examples. Grammars, theories and rules can wait. Take an algorithm, get data, apply the algorithm, interpret the results, repeat with another dataset. Little children do not learn a new language by learning grammar and rules. In fact they learn everything by watching and observing.

If you want to learn to swim, just go to a swimming pool rather than searching a “how to” video on YouTube, right? The whole world of machine learning is about examples – train a model with lots & lots of examples and then let the model solve a real problem based on what it has learned from those examples. Humans have figured out that this is the most efficient approach for machines to learn rather than giving them rules (because every rule has an exception). This is how human’s learning process should also work more efficiently. Once you dive into the technical analytical part with whatever tools/resources you have and you will automatically get interested in the mechanics of it, learn theories, while finding answers to questions you are curious about.

To be clear, this is not to say theories are more or less important. This is just to say that – between theory and practice – a learner should start with practice, theory will just fall in line in the process.