Avoid these pitfalls when getting started in data science
In me lately blog and video, I’ve received several messages from people looking for advice on learning data science. Although I can provide them with a sufficient roadmap, I believe it is equally important to highlight common mistakes that beginners often make.
In this article, I would like to outline six mistakes that, in my opinion, are commonly made by novice data scientists when trying to break into the field. If you can avoid even one of these pitfalls, you’ll be successful and do better than I did in the beginning.
All modern Python packages abstract away the need for mathematics, so many people today think that no knowledge of mathematics is necessary.never actually executes error backpropagation method Do it manually or build a decision tree from scratch (but it’s a lot of fun to try!).
It’s easy to take this for granted and avoid learning the background theory behind the algorithms, but this is dangerous and is not recommended.
Sure, you can build a neural network with a few lines of code in PyTorch, but what if you have a strange prediction and need to debug it? What do you do when someone asks you what the confidence interval is for the predicted output?
These questions and scenarios come up more often than you might think, but you can answer them if you have a solid understanding of the underlying mathematics.
I can empathize with the fact that math can be scary and not everyone may be good at it. However, the mathematics required for most data science roles is not at master’s or PhD level. This is typically what is taught in the second semester and first year of university for most of his STEM subjects.
STEM = Science, Technology, Engineering, Mathematics
I have a previous post detailing the math you should learn and the courses I recommend.