Random Noise #3 — Decision Intelligence
curating the unwritten world of analytics
Want these updates delivered to your inbox? Subscribe here.
“Mine everything for inspiration but don’t take yourself too seriously.”
Decision Intelligence & Data Science
Cassie Kozyrkov is Chief Decision Intelligence officer at Google Cloud. Last week we featured her talk on applied machine learning at the Tech Open Air conference and this week we wanted to dig deeper. Cassie hits on a lot of topics in this wide-ranging interview with Hugo Bowne-Anderson. The whole discussion is worth listening to…lots of threads to pull on. One subject that has been particularly interesting to me is the differences we see in organizational models for data science within businesses. Big ideas below…
Big Ideas
5 Organizational Models for Data Science
Centralized Team
- Doesn’t work in small organizations
- Associated with high-value business tasks 👍🏻
- Data scientists have low contact with decision function
Embedded with engineering — “just be useful”
- Influential to engineering
- Ambiguous role — “are you just an average programmer?” 😕
- No shelter from politics
Decision Support — attached to a leader
- Clear direction and protected value
- Possible dilution of data science skills — working on lots of other stuff!
Full-stack
- Avoids high collaboration costs
- Tough and expensive to find the unicorns 🦄💰
- Need expertise in multiple areas or may send org down rabbit holes 🐇
Decision intelligence
- Broad, light touch of analytics going on all the time — explore over exploit attitude
- Centralized matchmaking with internal groups + staff for heavy lifting
- Decision-makers own the dive and diagnose what is truly useful to the business.
Careful not to hire “a bunch of professors.”
Listen: iTunes | Soundcloud
Designing AI to Make Decisions — Kathryn Hume of integrate.ai is great at articulating AI in a way it becomes approachable. “…when we have a machine that’s diagnosing cancer, it’s the accumulation of thousands of people’s judgments represented in a mathematical formula. That’s it. It’s not an alien intelligence.” <- nice way to put it!
Listen: iTunes | Soundcloud
DataOps in Action — Shakeeb Akhter, head of Analytics at Northwestern Medicine, discusses DataOps. Shakeeb thinks of DataOps as a discipline that brings together all the different skillsets that are required on a data team to build the total analytic solution needed by the enterprise (engineering, architecture, analytics, data science and software dev). This is another great perspective on data team building…I like the concept.
Listen: iTunes | Soundcloud
How Data Powers the Orlando Magic — Jay Riola is the Senior VP of Strategy & Analytics for the Orlando Magic 🏀. Most of the attention for analytics in sports goes to player/game evaluation. Jay focuses on the business side of things — retention, pricing, schedule optimization, etc.. Fascinating stuff!
Listen: iTunes
Have something we should share? Let us know! randomnoise@adaptablelabs.com
Not subscribed to our newsletter? We’ll send the content directly to you. Sign up here.
