on 04-01-2013 5:58 PM
I have a composite forecast (100% each) consisting of 5 univariate models that should select the best model based on MAD. I have a product that has only 1 historical period of actuals in the last 36 months. I would have expected the Auto Model 2 to forecast "zero" but it has recommended a constant forecast of the same value as the 1 historical period for the next 24 months.
Is there a parameter that should be modified in order to improve the forecast. Any help on optimizing the parameters would also be very helpful. I have yet to hear of a great success story with Auto Model 2.
Thank you.
Hi,
if you ever hear a great success story about Auto.Model 2, you'll have all the right to be suspicious. You are more likely to find funny results from it (among many reasonably correct results, of course).
DP is just not designed for automatic forecast.
good luck,
J.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi,
You need at least 3 historical periods and 2 seasonal cycles (if valid) to initiate the model. Since the model is not initialized, the procedure used copy history model i.e. historical data used as forecast data for the entire forecast horizon.
Thanks,
Rajesh
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
User | Count |
---|---|
15 | |
4 | |
3 | |
2 | |
1 | |
1 | |
1 | |
1 | |
1 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.