sql server data mining: Hi, I have two questions. The first one: is there a way to have hierarchical input features in a model? For example, a location feature that would be city/state/country/continent. I'm aware of the nested cases but I'm not sure if they are equivalent. In the few examples I have seen the nested cases seem to represent some kind of flattened 1-to-many relationship in the data.

The other question is: how can I perform ranking? For example, if I'm doing estimation with decision trees or regression and I wanted the best n predicted outputs with higher case probability. Is that feasible with DM for SQL Server?

thanks,
Gustavo Frederico

[quoted text, click to view]

You are right, nested tables are useful for 1-m relationships. You could
model natural hierarchies with nested tables, although I cannot understand
at the moment what would you like to achieve.

[quoted text, click to view]

Yes, you can use DMX Select statement with some nice DMX functions like
PredictProbability.

--
Dejan Sarka
http://blogs.solidq.com/EN/dsarka/