Can you implement nearest-neighbor classification in Clementine?  Has anyone done it?

Clementine offers a few methods to cluster the data, K-means being a popular choice.

There is also a scaleable SPSS method called 'TwoStep', and a simple Kohonen implementation that can both cluster the data well.  Kohonen offers different visualisation options, but uses far more memory to build the model (so you cannot often build model using large samples).

Clementine's K-means node also automatically scales categorical inputs so these can be used (not just numerical data).

Cheers

Tim

But is it possible to use nearest neighbor as a classification tool rather than just a clustering tool in Clementine?

What are you trying to accomplish? 

With more information forum members might be able to suggest options. I'm tempted to reply with a succinct "yes"...  I have used Clementine's K-means to classify outcomes (such as fraud).

Tim