Noise, Deviation, Anomaly, & Mistake
It’s an exciting thing in science when you find something you didn’t set out to discover! Over the summer research period, I set out to better understand how engineers reason about variability. In the process of writing grants and designing interview protocols to elicit engineers’ thoughts, I had to do a lot of thinking on how variability “ought” to be defined. To that end, I developed a two-axis framework to describe variability.
These two axes lead to four disjoint “flavors” of uncertainty: noise, deviation, anomaly, and mistake. I’ll be teaching these ideas in Data Science this fall, and am currently operationalizing this framework as an interview protocol.
If you’d like to learn more, take a look at this draft chapter from a book I’m writing on modeling under uncertainty.