Orange Forum • View topic - Weighted association rules?

Weighted association rules?

A place to ask questions about methods in Orange and how they are used and other general support.

Weighted association rules?

Postby Jeff » Tue Jan 20, 2009 0:19

Hi, I am looking at Orange basket format for association rule mining as a simpler, speedier alternative to WEKA. I am impressed so far.

The docs say: "If examples are weighted, weight can be passed as an additional argument to call operator." How do I do this?

I use:
Code: Select all
rules = orange.AssociationRulesSparseInducer(data, data.weight, support = 0.5)

and get an error. (It works fine without data.weight.) What's wrong?

Thanks,
Jeff

Postby Janez » Sun Mar 01, 2009 19:10

Jeff, data.weight() (with parentheses) is a function which returns the total weight of examples in the table. AssociationRulesSparseInduces expects an ID of an attribute for weight. You can read about weights in http://www.ailab.si/orange/doc/reference/Example.htm.

(By the way, adding parentheses does something even worse: it crashes Orange, which is probably related to that id has to be a negative number. We'll fix that.)


Return to Questions & Support



cron