You’re Gonna Need Some Math!
As I’ve written about previously, at some point every aspiring data scientist has to learn some math for machine learning. To be blunt, the more serious you are about data science, the more math you’ll need to learn. If you have a strong math background, this is likely to little issue. In my case, I’ve had to relearn much of the mathematics (note – I’m not done yet!) that I took at a university as my professional life had allowed my math skills to atrophy.
Based on my experience teaching our bootcamp there is also a group of aspiring data scientists that fall into a category where their formal math training needs to be augmented. For example, we have many students that come from marketing backgrounds where, for example, studying linear algebra was never a requirement.
What Math Data Scientists Need
Forms of the question “what math do I need for data science” and “what math do I need for machine learning” are popular on sites like Quora. I would encourage all aspiring data scientists to perform their own research on this subject and not to take my post as gospel. However, as I often get asked for my opinion on what math aspiring data scientists need to know/study, I will provide my own list:
- Basic statistics and probability (e.g., normal and student’s t distributions, confidence intervals, t-tests of significance, p-values, etc.).
- Linear algebra (e.g., eigenvectors)
- Single variable calculus (e.g., minimization/maximization using derivatives).
- Multivariate calculus (e.g., minimization/maximization with gradients).
Please note that the above is not an exhaustive list. To be honest, you likely can never know enough math to help you as a data scientist. What I would argue is the above list represents the 80/20 rule – the 20% of math that you will use 80% of the time as a practicing data scientist.