Data science is a challenging field to grasp. Not only is it broad in terms of the topics you need to know, but it also requires a lot of depth in those topics. I’ve been teaching data science for some time now, I’ve seen these mistakes from my students, and I also made these mistakes after I went through a bootcamp myself.
1 Applying deep learning techniques to every single problem.
Deep learning is very powerful, and it’s very cool to say “I used deep learning” on a project. However, not every project requires deep learning. In my lectures, an analogy I always mention on always using deep learning is similar to “using a jackhammer to hammer a nail down.” There are many projects and use cases where a simple linear regression or another classical machine learning method can solve the problem.
2 Not spending enough time revising SQL.
No one is going to query your data for you; you need to query your own data. 9/10 times those data points are on a relational database that requires SQL to gather. Most bootcamps do cover SQL; however, after an intensive few months of learning. It is often an overlooked subject. I know I did because I fucked up a few whiteboarding questions after I graduated.
3 Not developing your product sense/business sense
A good portion of data science projects is always tied to a business objective.
For example, it can be increasing the underlying revenue by A/B testing a recommender engine to improve online shopping retention.
Or developing a classifier to help identify potential invoice fraud to speed up internal identification process.
In day-to-day work, a data scientist should have a good grasp of business sense and be able to communicate that understanding.