Estimating the unreliability of extrapolations

[Cinitiator]

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.

 

[Stephen Tashi]

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.

 

[chiro]

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?