Showing posts with label clusterdistance. Show all posts
Showing posts with label clusterdistance. Show all posts

Friday, February 10, 2012

ClusterDistance() of Zero for All Cases

I recently added a nested table to a model that I had been using for a while. I noticed that after I added the nested table that the ClusterDistance() function returned 0 for every case. I went ahead and changed some of the keys for the nested table records so that the values would show up as missing and now the cases with a missing value have a non-zero ClusterDistance() value. Can anyone help me understand why this may be happening?

Thanks.

How many clusters were in the model? And what is the result of the Cluster() function for each case?

Can you post the cluster content? (SELECT FLATTENED * FROM ModelName.CONTENT)

Thanks

-Jamie

ClusterDistance value

Can somebody please define the result of the ClusterDistance() function for me. BOL and MSDN are vague about what the output means and I am getting some cases that have a value of zero and sometimes there are more than one per cluster.

Thanks in advance,

Dan Meyers

I believe the clusterdistance() is simply 1-clusterprobability(). Are you saying you are getting more that one "0" value of cluster distance for different clusters on a single case?|||No. I am getting more than one case per cluster that has a ClusterDistance() value of "0".|||That's ok - it's an issue if you get more than one cluster per case with a ClusterDistance() of 0. Imagine if you have two identical cases - it would make sense that they have the same ClusterDistance(), would it not?|||

It does make sense. I just wanted to make sure that I was understanding everything correctly and I was not sure because the documentation is a bit vague in its definition of the function.

Thanks Jamie.