Ideal level of confidence/certainty

Dason

Ambassador to the humans
#21
As mentioned previously you really need to have some sort of cost/benefit information to do a cost/benefit type of analysis. You know - the specifics I was asking you to provide. Not sure why you're being so stubborn on that topic.

In all of your hypotheticals there really isn't any sort of optimization to be done because you haven't included any cost or benefit information into the problems. If we are just working with probabilities there really isn't anything to optimize.


There was no appeal to ridicule. There was no red herring. I've literally been trying to help and get you to define an actual question that can be answered. And starting with a specific example with actual numbers makes it easier to work out and follow the logic for all involved.

Best of luck.
 
#22
Moving on.

Solving for chanceToBeWrong:
tries = log(chanceToBeWrong)/log(chanceFailureEachTry)
chanceToBeWrong = chanceFailureEachTry^tries

test.png

The solution set is a line. So we can represent it in standard line format, y=mx+b

Any two points will do, so let's take [ (1, 0.99^1), (2, 0.99^2) ] and find the slope m
slope m= (change in y) / (change in x)
m = (0.99^2 - 0.99^1) / (2 - 1)
m = -0.0099

Putting the slope and point1 into the point-slope formula
y - y1 = m(x - x1)
y - 0.99 = -0.0099 (x - 1)
y = -0.0099 (x - 1 ) + 0.99
y = -0.0099 x + 0.9999
 
Last edited: