One way to learn anything quickly is to constantly apprentice yourself to people better than you at what you’re trying to pick up. This includes statistics and machine learning. It seems obvious, but people don’t do this enough. There are three major benefits to apprenticing yourself: standards, learning path, and direct information transmission.
Benefit 1: Standards
If you apprentice yourself to someone good, you learn their standards. In statistics/ML, there are several places you need standards. Firstly, it’s easy to gloss over the math/technical details of methods you’re using. When you work with someone good, they may grill you on ‘how does this work’ and keep drilling on details until either they understand it (in which case you do too), or both of you realize that you don’t ‘really’ get it and you have to get back to studying it. You also learn what kind of standards are sufficient for showing that an implementation is working, and what kind of standards are sufficient for showing that one method is better than another. Finally, many models come with pitfalls and sets of assumptions. If someone points out that you’re making assumptions that you aren’t aware of, this teaches you to think more carefully about assumptions.
Benefit 2: Learning Path
Often there is an order in which you should learn things. For instance, for undergraduate math/stat fundamentals it’s calculus->analysis->probability I-> math stat I and II, and somewhere before math stat II you want linear algebra. For well known curriculum items, you can find it online, but once you get to modern statistics or machine learning, it can be very difficult to figure out what you need to know to read a paper/understand a method well. Someone with experience can tell you read a, b, and c. Then when you get to d you understand it well. Without this, to be able to pick up the majority of things (including theory papers) on your own deeply, you’ll need training equivalent to the top statistics departments.
Benefit 3: Direct Information Transmission
Talking to an expert helps. Their knowledge permeates to you. They might tell you about recent approaches to a problem, a new software toolkit that makes your life easier, or teach you a new math trick. Further, in line with both of the above, you become more aware of what you don’t know.