Estimating the unreliability of extrapolations

  • Thread starter Cinitiator
  • Start date
  • #1
69
0

Main Question or Discussion Point

Are there any methods for estimating the unreliability of extrapolations? Obviously doing so is highly unreliable itself. However, I'm sure there are some factors, such as for how long a given trend persisted - if it persisted for a very long time, the future short-term extrapolations based on it will probably be rather reliable.
 

Answers and Replies

  • #2
Stephen Tashi
Science Advisor
6,924
1,182
Are there any methods for estimating the unreliability of extrapolations?.
It sounds like you are hoping to find some mathematics that does this without getting into a lot of detailed mathematical modeling of the phenomena that is being predicted or having a lot of data to test a prediction method against historical data. if that's what you want, you're out of luck. There isn't' any such universal mathematical method.
 
  • #3
chiro
Science Advisor
4,790
131
Hey Cinitiator.

There are methods that involve the variance of the model as well as the length of the interval over which the models data region is (to generate the fitted model) that give an error in terms of a variance but this is for a particular class of models (like a sub-class of linear regressions).

It's basically a conditional standard error of a y or y_bar given a particular x.

Have you seen this statistics?
 

Related Threads for: Estimating the unreliability of extrapolations

Replies
0
Views
478
Replies
4
Views
6K
  • Last Post
Replies
2
Views
1K
  • Last Post
Replies
1
Views
2K
  • Last Post
Replies
2
Views
4K
  • Last Post
Replies
1
Views
5K
Replies
1
Views
6K
Top