How to Be a(n Even Better!) Data Scientist in 2018

Data Basics Alivia Smith

It’s that time of year again: time to make (and hopefully keep this time!) your new year’s resolutions. If one of yours is to be a better data scientist, you’re in luck. We asked talented data scientists coming from different backgrounds and experiences what their advice would be for becoming a better data scientist.

sparkly banner how to be an even better data scientist in 2018

 

Their feedback can be summed up into these four categories (or you can download the full PDF with direct quotes from all the star data scientists):

  • Start Simple (and sometimes stay simple): It’s fundamental that you don’t give into the hype and build a complicated project with the latest tech from the get go. Start with basic data preparation and a basic model, and add to that little by little, testing that everything you add outperforms your first basic project. That way you know what you’re building is solid and not overly complicated.

quote from data scientist horizo and small animated headshot
  • Build for Production: To bring value, you have to build projects that aren’t one-off but rather that are built to run in production. This means working on data that matches your live data and can be added to your infrastructure. This also means reorganizing your work to focus on evaluation metrics even after your project is running in production to make sure what you built is right.
  • Keep Learning: Data science is, well, a science. So there are innovations every day, and it’s important to keep an eye on interesting papers so you can incorporate them into your job. It’s also important to look into methods in fields on the outskirts of data science - stats, biology, epidemiology. There are techniques there that you can transfer to your everyday projects as well.

quote from data scientist jorie and small animated headshot
  • Think About the Team: Build your code so it can be maintained by someone other than you, and know the basics about the ecosystem you’re working on to be more independent from your data engineer. Keep in touch with your business people - they know what the data means, and, more importantly, they’ll have to use your work in the end!

You May Also Like

Fine-Tuning a Model (In Plain English!)

Read More

How to Reach the Apex of Data Preparation

Read More

How to Address Churn With Predictive Analytics

Read More

What Is MLOps?

Read More