Wednesday, February 28, 2018
And by the way, on TA
By the way:
I'm taking spatial stats, and we're doing things like kernel estimates and Moran's I and so on.
Basically, with spatial data, you can generate these sorts of things to create maps that easily illustrate spatial variation in data.
One of my teachers was saying that he's been able to publish all sorts of stuff with a positive response just by doing a kernel analysis of a point pattern. It's a trivial exercise from a spatial stats perspective, but people in academia feel they get an awful lot of information from a picture.
I was thinking about this. Humans' brains are high-level visual computers: you can see how much so by looking at how fucking hard it is to train a computer to recognize a face, which we can do instantly.
And so, if you can present data in a visual pattern, you'll see what's subjectively a vast amount of info in that picture, that you'd never see in a list of numbers.
TA, then, is just applying simple heuristics to visual data, in order to access extra information visually.
And that's exactly why it looks like Tarot card reading, even when done properly.