Predictive analytics is gaining momentum and thanks to a number of easy-to-use tools, it is easier and easier to produce reasonably reliable forecasts. However, every forecast in the future is to a degree a shot in the dark. Perhaps not that bad, but it certainly needs qualifying.
This tutorial assumes that you have already generated your forecasts. It also assumes that you have conducted your error measurements and decided which one of the several models is the best to use. What you definitely need to do now is to create a prediction interval for your forecasts.
In this tutorial we revisit the rationale why we measure prediction interval the way we do and, more importantly, how do we account for the increase in uncertainty as we extrapolate our forecasts further into the future. To read the paper, just go here, and if you would like to see the Excel sheet, click here.