So last time we talked about decision trees and how can they leave to a certain outcome. And today we can talk about what final outcomes the decision tree can give us. So like if Randy finds himself in a good dimension, he can more likely find Anton at a computer shop. Or if he’s in a bad dimension, he’s more likely to find Anton in a place other than a computer shop. Like more in a place where big, bad Anton is likely doing big, bad stuff. So Randy assigns different probabilities to those two places, depending on what dimension he’s in. Like this.
Or … if we go back to fantasy casting for a moment and lets say I have two actors in mind for Anton and two actors in mind for Randy, and I think that Anton 1 would be more suited opposite Randy 1 and Anton 2 would be more suited opposite Randy 2. So if, say, the studio chose Anton 1, we might assign a probability of them choosing Randy 1 equal to 0.85 and of them choosing Randy 2 equal to 0.15 and vice versa if they chose Anton 2. Got all that? Eh, that’s okay. Besides we have a long, long way to see Order of The Dimensions as a hit movie if Order of The Dimensions ever becomes a hit movie. I know who I’d like to produce already though!
Which looks like this …
And how can they relate to the beloved characters in our story?
Well, lets say, Jane calls Tina since Randy is out of town on a major FBI mission or whatever and the kids are at a sleepover and asks if Tina wants to come over to see a movie.
Tina can say yes or no. If she says yes, Jane can for example ask her if she wants to get some Chipotle or something else. Well, Tina says Chipotle … duh!
So then Jane asks if she wants to see Order of the Dimensions or something else.
Well, Tina says … hey! What do you mean something else?
But anyway, Jane asks 6:30 or 7? And so forth … and I promise that Tina’s decisions are much more exciting in the books. Which, again, you can find here.
So that’s today’s lesson but next time we’ll extend the concepts of decision trees that could help, say, Randy prediction in which dimension Anton is in. Which is much more exciting that Randy choosing between a vanilla bean frapp and a double skim caramel macchiato — again, I promise!
So also with imputation models or any kind of model, really, we can adjust for different factors, factors besides the main one we want to look at. And such factors are called covariates. Kind of like Tina’s marital status can also be adjusted for her employment status or the weather of the city she’s in in January or her proximity to a Chipotle can further help us determine which dimension she’s in. That’s right one of the covariates change and our assumption of which realm she’s in can also change. Or again if I were to replace Krasinski as Randy Lipinski in my fantasy cast with …
my entire movie could change! Or else, I could also get four movies deals … and what? You’re right … you’re right. Maybe just three movies with Jason Segal and Justin Chatwin as Randy in two of them. Yes, I’ll stop.
Well, with Easter around the corner and work picking up, I’m taking a small break this week and leaving you with this image.