12. Frontiers

In this chapter we turn to two areas of data science that, while not exactly new (as we’ll see), are undergoing particularly exciting developments: reinforcement learning (RL) and natural language processing (NLP). We will give you just a flavor of how data scientists think about the world in terms of RL or NLP problems, as well as some intuition and practice with a classic problem in each.

As we go, notice that while we are using code and packages that we haven’t explicitly covered in the book so far, we hope you are pleasantly surprised by how much of the code you can already read and implement even if it’s technically brand new (though you’ll see some old friends in there as well!). In addition, as carry out types of analyses that are new compared to what we’ve been doing, we also hope you’re able to notice that you’re able to translate the world into RL and NLP data science problems relatively quickly compared to, perhaps, how it felt when you started this book.

This, of course, is at the heart of thinking like a data scientist: memorizing formulas and lines of code can be handy, and it can feel good to have a list of algorithms and packages you know how to use. But knowing how to translate novel questions in the world into data science problems, how to figure out how to use new code to iplement those translations, and how to thoughtfully adapt both to the particular needs of your dataset, the phenomena you want to better understand, and your own curiosities and goals – now that’s where the fun (and power) really starts.

Welcome to the next level :).