prediction interval vs confidence interval terminology

[brshallo]

"confidence interval" as used in {modeltime} refers to "prediction interval" in the sense that Hyndman means in The difference between prediction intervals and confidence intervals, correct?

Reference for modeltime_forecast seems to make this clear:

An 80% confidence interval estimates a normal (Gaussian distribution) that assumes that 80% of the future data will fall within the upper and lower confidence limits.

Am wondering if it may be helpful to sneak in a few mentions to "prediction intervals" in documentation (for those of us on the look-out for that type of terminology...)

[AlbertoAlmuinha]

Hi @brshallo ,

That is correct. We always return prediction intervals but not confidence intervals. Maybe we should change the documentation to refer to these concepts and make it easier for everyone to be clear about what we are working with. Let's see what @mdancho84 thinks about this.

Regards,

[spsanderson]

I would second this

[mabuimo]

I agree that prediction intervals and not confidence intervals is the term that should be used.