Confidence Interval

If we are using linear regression, you must realize that the line of best fit will never be perfect. Errors are unavoidable or else you might be Nate Silver. We will never have an accurate predictive model. We tend to the confidence interval to comfort us that the result will fall within a certain range. A confidence interval is more often than not a 95% confidence interval where we have 2.5% percents at tails. Any information not within the field, and in the tails, indicate that the results are questionable.

Prediction Interval

The data we are collecting will never be accurate. The questionnaire might have reported information inaccurate, or a data analyst could have queried incorrect data. There’s nothing that prevents these mistakes. Thus, generally, the distance in the prediction interval is larger than the confidence interval.