sql server data mining: Null values -- sql server data mining

Null values

Cesar
4/27/2004 7:05:23 PM

sql server data mining: I have another question related to the NULL IGNORE clause
one.
Do you know how the Null values are treated in a
clustering relational data mining model? Can the user
control this behaviour?

For example, I have the following table:

Car Model Feature 1 Feature 2
Feature 3 .....
Accord 1989 0.5 1
NULL
Accord 1988 NULL 100
NULL
....

And the problem is that I'm getting cluster members that
don't have any common column with the other members of
the cluster and their probability is almost 1. I think
those strange members are being included because of their
missing values (. NULL INFORMATIVE: Data mining
algorithm will model the missing state.). I want to
ignore null values and make new clusters for those
members or assign them a very low probability.

Do you have any suggestion?

Re: Null values

Peter Kim [MS]
4/28/2004 11:02:48 PM

Yes, we model missing as informative. You might want to try to map NULLs to
the mean value of the column to see if you can get what you want.

--
Peter Kim
This posting is provided "AS IS" with no warranties, and confers no rights.

[quoted text, click to view]