DMX to read cluster characteristics

Hi,

Can you send me or point me to a DMX code to read cluster characteristics programatically ?

Thanks,

The information displayed in the Cluster Characteristics viewer is available through a stored procedure call.

CALL System.Microsoft.AnalysisServices.System.DataMining.Clustering.GetClusterCharacteristics('Table1_CL','001',0.0005)

The first parameter is the model name, the second is the NODE_UNIQUE_NAME for the cluster whose characteristics are to be rendered and the last parameter is a threshold.

You can get the list of all clusters and their associated NODE_UNIQUE_NAME with a query like:

SELECT NODE_UNIQUE_NAME, NODE_CAPTION, NODE_DESCRIPTION, NODE_TYPE FROM [Table1_CL].CONTENT WHERE NODE_TYPE=5

(NODE_TYPE=5 identifies a Cluster node)

Hope this helps

|||

Hi Bogdan,

This worked greart. Thanks.

Can you also let me know the online material or book where I can get some more information of this kind about DMX to access information from SSAS DataMining model ? We are looking forward to write more of such codes to get information from the mining model programatically.

Thanks,

Vikas

|||

There isn't a lot of documentation on the stored procedures that retrieve summarized/derived information from the algorithms. All of the information is derived from the content (i.e. SELECT * FROM <model>.CONTENT). You can use SQL Profiler to inspect the queries that are sent from the viewers to the server to access the syntax and how the sprocs are called. Also, you can look at the sample thin client viewer code that ships with the product to see the characteristic and discrimination calls against clustering and naive bayes models.

To get a better idea of how the content looks per algorithm, you can download a sample plug-in viewer we wrote from http://www.sqlserverdatamining.com/DMCommunity/Downloads/Links_LinkRedirector.aspx?id=1349. This viewer provides a nicer interface for navigating generic content, decoding node types and value types and such.

|||

Here's one way of getting the characteristics that distinguish a cluster. Feedback welcome:

http://www.codeplex.com/ASStoredProcedures/Wiki/View.aspx?title=ClusterNaming&referringTitle=Home