When To Use A Nested Table?

Why would someone choose to model a problem with a case and nested table versus just using a case tables with multiple rows?

For example, a simple customer and product scenario where you could say:



Mining Model DT_CustProd
(
[Id] ,
[Gender] ,
[Age]
[Products] Predict
(
[ProductName] ,
[Quantity]
)
)



But you could also model as:



Mining Model DT_CustProd
(
[Id] ,
[Gender] ,
[Age]
[ProductName] Predict,
[Quantity] Predict
)



and just insert multiple rows per Id/Gender/Age? Is there any rule of thumb to use here when deciding to model with a nested table?


I understand that a nested table essentially pivots the data (http://forums.microsoft.com/MSDN/ShowPost.aspx?PostID=408840&SiteID=1) and the data becomes an attribute, but how do you know if you want that, or you want the data as a value of an attribute?