on 07-19-2012 5:13 PM
Hi Gurus,
we are currently implementing the DP for a startup company. the business was started 2 months ago. we have altogether 100 products and we would like to know how the forecasting could be used in a scenario where there is no historical data or very short historical data say in this case only for 2 months history..
could you please share your ideas...
Best Regards,
Hello Shreeyath,
For your current situation, You can use
- Moving Average
- Weighted Moving Average
- Constant Model with 1st Order Exp Smoothing
- Linear Regression
Hope this helps,
Hari Vomkarey
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi Hari,
Thank you, but since i have only 2 months historical data, could i still use these models, i know that if the system does not find the adequate history(say continuously for 5-6months), it automatically chooses the constant model..is nt it?
what happens if i do the foreasting with weeks data i.e. i will split the history into weeks instead of months and do the forecasting..is this the correct approach..pls share ur ideas..
Best Regards,
Yes, you could use these models with only 2 months of History,
Moving Average and Weighted moving Average should work without any errors. If you use weighted moving average (strategy 14), try creating weighting profiles for 2 months in the Univariate forecast profile. If you get any errors, like -insufficient history..check and maintain master forecast profiles with just 2 months as periods for History horizon. As you develop sufficient history, you can change this to 3 months, 4 months ,...so on.
Forecasting on weeks , technically will give you more history for forecasting. But this decision of using which planning levels (weekly/monthly) should driven by business/planning needs of your company than merely being able to get forecast numbers.
Hope this helps,
Hari
User | Count |
---|---|
7 | |
4 | |
3 | |
2 | |
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.