Meeting with Stephen this morning, will see what his ideas are which will affect what I do this week. Also, chat with Liam this week which will, once again, affect things.
The general model is the fishbone diagram. For this week I’ll work on perturbation analysis, so moving variables by 1 standard deviation from the expected value and seeing effect on outcomes.
Because each input has its own specific domain range (if its an rv) then its a further case that inputs should be considered random variables. Or just have a special type of input?
Am I pestering Mark too much?
What I have at the moment is moving within a ten percent range each input. A further refinement would be to take a more ‘educated’ guess on this range through assumptions etc. Given this, is it a worthwhile output?
What could be considered a ‘finished’ perturbation analysis (should I look to research)?
Lindley defines the balls in the urn as a standard like a unit of length, trying to intuit how far to extend this analogy based on what I think probability (a mathematical model) is so far.