Data Science: A Mindset for Productivity
Today, I had the pleasure of delivering the data science keynote at the West Coast CTO Summit organized by Ronin Labs. I was in great company -- fellow speakers included Raffi Krikorian, Cathy Polinsky, and other technology executives sharing their wisdom to help all of us build and manage successful engineering teams.
My talk, "Data Science: A Mindset for Productivity", emphasized problem formulation over problem solving. The tl;dr is that "The most important part of data science is picking the right problem and figuring out how to frame it."
I reiterated the "three ex's":
- Explain: Iterate using explainable models.
- Express: Model your utility and inputs.
- Experiment: Optimize for speed of learning.
I'm thrilled by the amount of attention that data science has received in the past several years. But I hope that we don't get so caught up in the technologies and algorithms that we neglect what matters most: solving the right problems.
Security - AI Product Leader | Ex-Amazon AI | Ex-MSFT | Northwestern Kellogg
9 年Good insights. One thing I would like to add is model simplicity. The reason I emphasis this is at the end of the day, the model needs to get adopted by business leaders and a data scientist needs to be able to explain the approach and results in a simple business language. This increases the adoption rate and the confidence with Data Science teams.
Query Understanding
9 年Thanks all. Hope video will be available soon. And apologies if some of my points were oversimplified -- sometimes it's hard to be nuanced in a keynote. :-)
Data Engineer @ Netflix | ex-Airtable, Airbnb, Facebook
9 年Is the video for the talk available?
Great! I wholeheartedly agree with most of this. One exception: I don't think "test one variable at a time" is necessarily good blanket advice. Design of Experiments is all about testing many things at a time efficiently.