0

I not sure to what field exactly this question belongs, but math/statistics seemed closest to me. So here we go:

It is a method of estimating a value that informally goes like this (bear with me). You pick two values. One that is clearly too low and another that is clearly too high. Then you try to bring values closer together by repeating the first step. When you found minimal too high and maximal too low values, that you can still define as too low/high, you stop and choose a value between and that is your estimated value.

I would like to know what is the official name of this method (if it has one) or if there is any similar or related mathematical/statistical estimation method?

zigzag
  • 101
  • 1

1 Answers1

0

If you take the midpoint between your high value and your low value, and then substitute that in for the new high or low (depending), that would be called a binary search. There may well be other names for it, but that's the one I'm used to.

Mitchell Kaplan
  • 515
  • Thanks! This one is close. Do you know if that method could be adjusted if not all values could be defined as too low or too high, but only assigned probability and confidence. Where values close to result have lowest probability/confidence. – zigzag Nov 06 '15 at 16:58
  • I don't really know. I think though, that if your probability was a smooth enough function that some standard numerical methods might work. Such as Newton's method. However in that case you're generally starting with a single "guess", not a high and a low. If you're thinking about this sort of thing, you might want to add a tag of Numerical-Methods to your question. – Mitchell Kaplan Nov 06 '15 at 18:15