How to prove my forecast is not bad?

[fish]

Hi all,

I'm verifying some forecasting methods with 36 months history data. Then I
make a 12-month sales forecast compared with real demand. My purpose is to
say that this method is precise enough to be a forecasting method in this
environment. Please take a look at this picture:
http://218.1.33.178/fish/s.jpg .
I don't know which statistics tools I should adopt. Please give me some
suggestions.

zh.y

 

[JE McGimpsey]

There's absolutely no way that anyone could even venture a guess as to
how good the model is with the data you've provided. Forecast validity
depends in large part on the volatility of the inputs for the time
period of interest (as opposed to the seasonality, which your model
shows). It's especially hard since you only included 11 months of data,
and the two ends have slopes of opposite signs - is there a cyclical
decline in October 2004? or is the prediction for unlimited growth?

What you're asking for can't reliably be answered in this newsgroup.