Click here to monitor SSC
SQLServerCentral is supported by Redgate
Log in  ::  Register  ::  Not logged in
Home       Members    Calendar    Who's On

Add to briefcase ««12

Help with Data-Mining \ Clustering question Expand / Collapse
Posted Wednesday, March 21, 2012 11:42 AM
Forum Newbie

Forum NewbieForum NewbieForum NewbieForum NewbieForum NewbieForum NewbieForum NewbieForum Newbie

Group: General Forum Members
Last Login: Wednesday, March 21, 2012 11:41 AM
Points: 6, Visits: 5
Carlos Bossy (3/21/2012)
I'm glad I could help so it's good to hear this has been useful.

With decision trees arriving at the same answer using multiple paths, the paths are not necessarily mutually exclusive. They might end up that way, but that would just be a random occurrence. For a tree of any complexity you will most likely have some overlap of variable values. In one path the node might ask 'Is age between 25-44' and in another it might say 'Is age > 39'.

I guess that's why I said you might not be happy. I got the impression you wanted a cleaner tree, with variable values occurring only in a single path. What you'll get instead for the multiple paths that lead to 100$-200$, path 1 might use age between 25-44 to make a decision, and path 2 might use age > 39 to make a decision. In fact, later in path 2 after satisfying age > 39, it might then ask is age between 50-65. If you are comfortable with this then you are good to go.

Thank you. It is deeply appreciated.
Post #1270385
« Prev Topic | Next Topic »

Add to briefcase ««12

Permissions Expand / Collapse