Andrew Gelman:
I suspect the words have different meanings in different contexts. In statistics, “prediction” is often used even when the result has already happened: that is, if you have a model, y = f(x) + error, then f(x) is said to be the “predicted” value. So you can have a predicted value, even for an observation you’ve seen. I’ve worked in an example using climate models where the term “hindcast” was used, because we were fitting a model from 40 years of data and then using it to predict values that had already occurred.
In common usage, I think “prediction” refers to predictions of the future. Consider the famous Niels Bohr quote, “Prediction is very difficult, especially about the future.” I assume he as being ironic and was assuming that prediction is always about the future. I can only assume forecasting is also about the future, given the “fore” at the beginning of the word. So I don’t really know the difference between prediction and forecasting.
I’ve never heard the term “collective forecasting,” but I’m sure that reflects my ignorance rather than anything else.
Previously: Apple dictionary on “predicting” vs “forecasting”
Andrew Gelman’s excellent blog
Andrew Gelman sometimes guest-blog at Nate Silver’s blog.
NEXT: Predicting = Forecasting –> Collective Forecasting = Collective Intelligence That Predicts
