It’s an exciting time to be a data scientist. There’s tremendous opportunity to rethink old approaches in both expected (technology companies / quant finance / etc) and unexpected places (like fashion!), and to incorporate algorithmic decision making in new domains. We’re still very early in this transformation in many industries.
Things are also moving quickly. The state-of-the-art is changing at seemingly ever increasing pace. It can be overwhelming to think about achieving some degree of mastery over this content – talk about a moving target! Don’t let this discourage you.
As with many things, it’s often best to start with the basics. A generalist, when equipped with a few basic but well-understood tools, is surprisingly effective. Time spent really engaging with topics like the following will seldom be regretted!
- Linear models and their modern extensions (GLMs, regularization)
- Frequentist and Bayesian inference and its applications to experiments (bonus: causal inference, selection bias problems)
- Optimization algorithms for fitting models (gradient methods, convex optimization)
- Good engineering practices (organizing code, writing tests, version control)
As you learn, don’t worry about the cutting edge. Focus on the basics and you’ll build a lasting foundation that will allow you to quickly reach the cutting edge of many different fields.