3 Things I would tell myself before I started my Data Science Bootcamp
- EDA and Statistics is taught for a reason
When I first started the Bootcamp I was surprised to see two whole months of Data Visualization & Exploration (EDA) and Statistics. Because I had the misconception that Data Science is all about Machine Learning. Although this phase was more fun than I expected, I never really understood why it was so important. As soon as we shifted into Machine Learning I got lazy and returned to my newbie strategy of throwing data at a model. I was able to get away with this without suffering too bad, mostly because my main focus shifted to doing Deep Learning with unstructured text data where EDA isn’t quite as important. However, my projects were still somewhat black-box and not the most interpretable. It wasn’t until I got on LinkedIn and read more about Data Science being used in real business scenarios that I saw the value of EDA and Statistics. I learned that EDA and Statistics are key to the Data Science Pipeline, not so much for the model building aspect, but more so to have a clear understanding of the data and be able to back our observations on a more factual basis. You could have the best model in the world, but if you’re unable to fully understand your data, have the ability to explain what is going on, and back up what you’re saying in a way that captures the stakeholder’s attention then they are very unlikely to buy in. If I could go back I would’ve kept the emphasis on statistics even into the Machine Learning phase.
2. Start building your network
During my bootcamp I was 100% focused on learning the technologies and while this allowed me to push through projects quickly, I graduated with very few connections which ultimately makes it harder to get a job. In general I feel that having a strong network is more beneficial in landing a job than memorizing all the Keras documentation. The job market is a competitive place and many employers end up filling roles through employee referrals because they stand out on top of the heaping pile of resumes. If I could go back I would have reached out to Data Scientists every day to start building connections early.
3. Seek to understand the domain too
We would end each phase, 3–4 week block, of the bootcamp with a data project. I remember some of my classmates did quite a bit of industry research alongside the technical component of their project, for example when we did a house prices prediction project many of my classmates did a good bit of research on real estate in the area (Seattle in this case). I didn’t think much at the time, but looking back I can see exactly why they did it! If this was a real project we’d have to pitch it in a way that people in the particular domain (Real Estate in this case) would understand, this is really hard to do without taking some time to at least have some context as to how the industry works.
If I could go back and talk to the me from 8 months ago that started this bootcamp I would tell myself to focus on Statistics and continue practicing it into the later phases, make connections with Data Scientists outside of the bootcamp, and take the time to fully understand the business context of each project I worked on. Let me know if this was helpful, if you have similar experiences, or maybe have a totally different experience.
