- by kq4ym, Data Doctor
- 10/9/2015 9:46:43 AM
A good example of being careful of our mind's tendency to be wired to think in a certain way depending on our past experiences, social norms, and expectations. Expecting to "name" a state we say Utah. A resident of the Texas where the county lies, may come up with the county as the answer A non-US person would come up with even a different answer.
- by Robert Allison, Blogger
- 9/30/2015 7:41:24 AM
I have mixed feelings about whether to accept analytics that are less than 100% correct ...
On the one hand, there are things like the 80/20 rule (with 20% of the effort, you can get 80% ... but to get the other 20% done it takes 80% of the total effort).
But then there's the way my buddy who used to teach engineering classes puts it -- he told his students that although making <100% on an assignment in school might be passing (for example, a 93% might be an A), that's not good enough in a real-world engineering assignment. In the real-world, you need to be 100% correct ... otherwise bridges & buildings will fall, and possibly kill someone.
- by Zimana, Blogger
- 9/30/2015 1:42:43 AM
I agree, Utah comes to mind. But that's how we consider looking at the United States. We look at states rather than counties. 50 states is easier than the thousand of counties, or even more towns. Hey, states is in the name United States, so we have reinforcement there as well.
- by Joe Stanganelli, Blogger
- 9/29/2015 5:56:30 PM
One of the other lessons of this, I think, however, is that data scientists are beginning to recognize that accuracy is but one axis to be concerned with in a lot of analytical applications, and sometimes not the be-all and end-all.
To most audiences outside of Missouri (and perhaps even inside Missouri), the shape is Utah-like enough. And you want automated tools that can return the right answer the vast majority of the time when only incomplete or even partly erroneous information is available. Most of the time, a decent automated tool will get the right answer.
Would it be best if 100% of the time your analytics were completely correct? Sure. And you are dead-on about the need to fight against assumptions despite data that indicates otherwise. But as organizations are discovering that they can get more done more effectively when a small amount of accuracy is sacrificed, there are other important lessons to now absorb as well.
- by Joe Stanganelli, Blogger
- 9/29/2015 5:49:52 PM
Yes, I saw it wasn't quite right for Utah myself, but it was the closest thing I could think of (along with New Mexico, upside down, and Colorado/Wyoming with a "hat") on the state or larger level.
- by rbaz, Data Doctor
- 9/28/2015 8:24:27 PM
I also would guess Utah since we tend to run first towards the familiar and Texas counties are far from familiar to me. Good example for urging to carefully scrutinize the data with as little assumption made as necessary.
- by PredictableChaos, Data Doctor
- 9/28/2015 5:41:27 PM
The shape isn't proportioned right for Utah.
The right hand side looks something like the eastern border of Wyoming + Colorado. But that idea doesn't work for the left hand side at all.
With the 'Texas' hint, I expected it to be a couple counties in Texas. Which ones? I have no idea.
Still wasn't expecting a Texas county in Missouri.
- by jmyerson2, Data Doctor
- 9/28/2015 4:19:14 PM
I would have chosen Utah as the best choice. But then when I scrolled down, it turned out to be a Texas county. While I am not familiar with counties in that state, I think it's a great way to educate people not to jump to conclusions too fast when they see or evaluate data.