Clustering your data on a sample and then appending clusters is a common theme
especially if you are in customer relations and marketing related divisions...

When it comes to appending clusters that you have calculated form a 20K sample and then you're going to "score" a few million clients you still need to download the data and use the append cluster...
Why don't we have an In-db append cluster instead, which will quicken the "distance based" scoring that append cluster does on SQLServer, Oracle or Teradata... |  |
Best