RE: DMX Content Query
- From: "Bogdan Crivat" <bogdanc@xxxxxxxxxxxxxxxxxxxx>
- Date: Thu, 14 Dec 2006 15:40:21 GMT
You could call the stored procedure used by the cluster viewers.
I'd suggest:
- start the Profiler
- launch the Cluster viewer
- choose the Discrimination panel and display discimination between one cluster and its complement
- extract from the Profiler the stored procedure call (CALL .. GetClusterCharacteristics(...))
- call that procedure in your code to get the most significant attributes for the current cluster.
Hope this helps
bogdan
.
hi all,
I want to write a content query against my Clutsering model...
I have figured out smt like this:
SELECT FLATTENED
(SELECT ATTRIBUTE_NAME, ATTRIBUTE_VALUE, [SUPPORT], [PROBABILITY]
FROM NODE_DISTRIBUTION ) AS d
FROM PersonelCluster1MM.CONTENT
WHERE Node_Type = 5 AND NODE_CAPTION = 'Cluster 1'
But, what I really want to do is to get a list of clusters together with the
most important attributes contributing to that cluster.
I tried to use 'top 10' and 'order by Probability', but it gives errors,
either says options are not allowed in sub-select, or smt like that...
Is there any idea or any resource explaining how to make content querires in
depth...
Thanks...
- References:
- DMX Content Query
- From: Erhan Erisken
- DMX Content Query
- Prev by Date: RE: Report Server and Data Mining Connection error
- Next by Date: Error using OPENROWSET
- Previous by thread: DMX Content Query
- Next by thread: Report Server and Data Mining Connection error
- Index(es):
Relevant Pages
|
