5 Things I learned as a Data Scientist
I recently completed 2 years of working in the industry as a data scientist. In this time, I’ve worked on research ideas, I deployed my first model to production and presented my work. Today, I would like to reflect on how far I’ve come, share some ideas I’ve learnt in my journey and hopefully start a discussion on what others have learned as well. Below are five takeaways from my past two years as a data scientist.
Lesson 1 — Clarify.
I used to be one of those that jumped into conclusions and made assumptions based on the facts presented to me. This resulted in many hours of wasted effort and reworking when all I had to do was clarify at the start. Ugh. Clarifying sometimes feels like an unnecessary exchange, but I learned how important it is to ask questions if you don’t understand something.
Lesson 2 — Document.
Document everything. Data Scientists run a lot of experiments and it can get so easy to lose track and forget to document. On many occasions, I have been lost in a maze of my own making which later led to a lot of frustration. You assume you’ll understand the code that you wrote a few months back, but you don’t. For people like me, if you’re not documenting more than you feel comfortable, you’re not documenting enough.
Lesson 3 — Always be Learning.
You might never know when those online courses you are taking come in handy. Read blogs. Follow data scientists on Twitter. Talk to friends. I like some of the blogs listed below:
https://www.allendowney.com/blog/
https://www.naftaliharris.com/blog/
https://www.analyticsvidhya.com/blog/
Of course there are many more great blogs out there, but I really like these. Let me know if you have a favorite blog that you think I should know about.
Lesson 4 — Be coachable and be open.
I’ve learned the importance of being flexible. The ability to be flexible and coachable can provide great learning opportunities. Being great at one specific thing is wonderful, but it’s also a double edged sword. If you’re not flexible you run the risk of stagnating ( unless you absolutely love doing the same thing over and over again ). I strive to be coachable and open minded so that I can tackle any kind of project that comes my way.
Lesson 5 — Keep it simple.
Everyone in the data science industry tells you to first understand the business problem, build the simplest solution and iterate. However, this was hard for me to follow. On many occasions, I’ve found myself struggling to explain why I chose a complex model. As an example, at one point, I built an elaborate solution which made use of multiple models. I felt great while experimenting and was so excited to share my creation with everyone. Unfortunately, the harsh truth was that this model was creating a ton of confusion and finally, I had to simplify.
I still experiment without abandon, but I’ve learned over time to critique my own work. When in doubt, keep it simple!
There are many more lessons I’ve learned along my journey, these are just a few I hope others may find useful as well.
What are some things you all have learned that might be applicable to others?
