What We’re Reading - 01/09/2020

Did you know you can now sign up for weekly-ish updates to our blog via email?

While I’m still working on the next post in our NFL Analysis series - this time on modeling Field Goal attempts - here’s what I’ve been reading (and watching) lately:

How to Succeed in Business by Bundling – and Unbundling

Over the holidays I reread a few older posts on business and strategy such as this interview with Mark Andreessen and the inimitable Jim Barksdale. Here is part 2, Don’t Play with Dead Snakes, and Other Management Advice

As someone who’s old enough to remember the first release of Mozilla and Netscape, I also really enjoyed this collection of quotes attributed to Jim Barksdale, some of which I’ll keep handy as future meeting material: JB Sayings

Data project checklist · fast.ai

Jeremy Howard with an exhaustive checklist for new data consulting clients. I will borrow steal liberally from this.

How To Speak by Patrick Winston - YouTube

Patrick Winston taught, amongst other subjects, a popular course on AI at MIT for years. His lecture on public speaking is well worth an hour of your time.

From the comments:

“We’re so lucky that this gem of a lecture was captured before he died. Now he can deliver this talk every year, just like he did before.”

Progress in the past decade « Statistical Modeling, Causal Inference, and Social Science

Andrew Gelman with a wrap up of his and his collaborators’ massive body of work over the last 10 years. I don’t think Andrew sleeps.

The Silent Bug I Find in Most Triggers - Brent Ozar Unlimited®

Brent Ozar is the unofficial ambassador of Microsoft SQL Server and runs a successful consulting and video-based training company. And like all good consultants, he mostly helps you find embarrassingly simple bugs like this one. (Brent is like the Bill Burr/Anthony Bourdain of SQL Server training and apparently pulls in north of $1 million a year in revenue. Good for him, the videos look like a ton of work.)

Wizard - our ML tool for interpretable, causal conversion predictions | better.engineering

Really interesting work at Better on conversion prediction, that also highlights the use of dbt models as great tools for feature engineering, something I’ve been advocating and doing in my own work.

Distribution Fitter 5000

This is a fun tool built with Streamlit that lets you fit a number of statistical distributions to your data.