Log in

View Full Version : Formula help


Th'Pusher
04-18-2013, 04:18 PM
Totally not applicable to the political forum, but, I am trying to evaluate the accuracy of a forecasting system we developed so I am using the mean absolute percentage deviation which is really simple =abs(demand-forecast)/demand. The issue I am having is that I can't necessarily control the amount of available product. For example, my forecast says we'll sell 100 units, but only 75 units get produced. All 75 units get sold. My MAPD says my forecast was 33% high which may or may not be the case since 100 units were not available to be sold.I have a similar metric that shows how production deviated from the forecast, but I am having difficulty quantifying the accuracy of the forecast. Any ideas?
CosmicCowboy
04-18-2013, 04:29 PM
Seems like product inquiries or backordered product orders would prove or disprove your forecast.
Th'Pusher
04-18-2013, 04:40 PM
Seems like product inquiries or backordered product orders would prove or disprove your forecast.

Yeah. These are commodities with easy substitutes so no such luck there.
CosmicCowboy
04-18-2013, 04:56 PM
Well, clearly production was sold out so you know demand was greater. If it's a commodity with easy substitutes I'm assuming the product isn't terribly expensive. Even if it is, you know there is demand for it. Produce 125 this run and see where you land. It's either inventory or sold. They are both money in the bank.
RandomGuy
04-18-2013, 06:44 PM
Totally not applicable to the political forum, but, I am trying to evaluate the accuracy of a forecasting system we developed so I am using the mean absolute percentage deviation which is really simple =abs(demand-forecast)/demand. The issue I am having is that I can't necessarily control the amount of available product. For example, my forecast says we'll sell 100 units, but only 75 units get produced. All 75 units get sold. My MAPD says my forecast was 33% high which may or may not be the case since 100 units were not available to be sold.I have a similar metric that shows how production deviated from the forecast, but I am having difficulty quantifying the accuracy of the forecast. Any ideas?

Hmm. You will need to expand on this a bit to get a good answer.

If you are trying to get at supply/demand stuff, you may want to look into economics a bit. They have fields of study that look at this kind of thing.

Will put some thinking into it at some point.