Developing a Forecasting App -- But with Variable Prices...

Question

Hi,
Would like to post an interesting question for the BPC community:  We're developing an application to do Sales Forecasting which will replace an app from a competitor.
The one tricky thing with it is that the prices for products can vary by customer, order size, and by month.  Sometimes the variations are quite significant, therefore we have to find a way to incorporate them into the application in some way.
Right now, our best idea is to enter the forecast by QUANTITY and then (maybe) have a whole separate dimension (?), to list out all sorts of different prices.
Has anyone seen other approaches in other Forecasting app's?
Thanks,

Answer 1

Signeddata is the only numeric field in the fact table where all values are stored.  You're confusing table structure with dimensional values.  The fact table is comprised of one column for each dimension in the application and a numeric field called signeddata.  So if you have 5 dimensions in your application there will be 6 columns in the fact table.  What the amount in signeddata represents is based on what dimensional values you have assigned to the record.  From the earlier discussion if you went with separate accounts (quantity, price, sales) then you'd have records in the fact table similar to:
Category, Time, Product, Customer, Account, Signeddata
Forecast, Jan.2008, Product x, Customer y, Quantity, 100
Forecast, Jan.2008, Product x, Customer y, Price, 5
Forecast, Jan.2008, Product x, Customer y, Sales, 500
If you loaded a record from your GL then it would look something like:
Category, Time, Product, Customer, Account, Signeddata
Actuals, Jan.2008, Product x, Customer y, Sales, 900
If you used the second layout discussed (with a data source dimension) then you'd have forecast records that look like:
Category, Time, Product, Customer, Account, DataSource, Signeddata
Forecast, Jan.2008, Product x, Customer y, Sales, Quantity, 100
Forecast, Jan.2008, Product x, Customer y, Sales, Price, 5
Forecast, Jan.2008, Product x, Customer y, Sales, Dollars, 500
Actuals, Jan.2008, Product x, Customer y, Sales, Dollars, 900