Open Mttbnchtt opened 5 years ago
Let us try to come up with an acceptable definition of forecasting. Here is my attempt: a process during which an agent uses historical data to make informed estimates of whether a specific event will happen or not.
Any suggestion @marieALaporte and @pbuttigieg?
I think forecasting does not need to be planned. For example, one may go out and come back to get an umbrella because, having observed a cloudy sky, forecasts that it will rain.
Forecasting is always planned, even if the plan arose in a few seconds in one's head.
I use historical data and event, which are not items in SDGIO. I think their meaning (or nature) is clear enough that this is ok.
They are out of scope for us too. We can add a request on the IAO tracker for something like historical data (very useful), but it may take them a while to generate it.
"event" is synonymous with process in most cases.
Forecasting does not need to make it explicitly the point or range of time in which an event is estimated to happen or not. For example, The Maltese soccer team will win the World Cup is a forecasting as well as It will win the World Cup within 10 years.
You had to go there didn't you ⚽️
I think we need a definition from a modelling perspective. Modellers and statisticians differentiate prediction and forecasting quite strictly in most cases. Here's a blog post on it, but please find a more reliable source like the OECD manuals.
Thanks for the references. Examining the sources, I realize that the issue is complicated. The first source does indeed distinguish between forecasting and prediction. The second source, remarkably (but not quite clearly), does not ("Forecasting and prediction‖ are often used synonymously in the customary sense of assessing the magnitude which a quantity will assume at some future point of time").
I also realized that the definition of data in IAO is wrong (or so I think). IAO says that data are statements. However, a name may be a data item. This makes it delicate to ask them to add historical data. I am not sure that I should pester ontologies asking for corrections more than what I am doing already.
In addition, I realize that a bunch of terms related to forecasting are probably useful to speaks about processes to respond to climate change. Examples are estimating and projecting.
As a result of these considerations, I need to think more about how to proceed wisely.
We should add forecasting and climate change forecasting as classes and link them to planned process to improve the forecasting of impacts of climate change.