1
$\begingroup$

The situation is as follows. There are 400 examples in the training set and 200 discrete classes (each class has two examples). There are a few thousand attributes.

When I run dimensionality reduction to 2D or 3D, I would like to see (optimally) 200 clusters of 2 points each - one cluster for each class. However in practice that's not the case.

The question is, how do you go about choosing the set of attributes that will give optimal results with clustering/classification (whether with dimensionality reduction or without it).

$\endgroup$
  • $\begingroup$ Can you provide any context about the type of data this is and why you only have 2 instances for each class (I'm just curious)? $\endgroup$ – Nick Apr 7 '12 at 0:50
1
$\begingroup$

Even if you would see "clusters" of 2 points each, it would not be statistically meaningful. You need more data. Two examples per class is just too little.

$\endgroup$
  • $\begingroup$ Two examples per class is what I have. I'm not looking for statistical meaningfulness, I'm looking for a good fit. $\endgroup$ – Nucular Mar 8 '12 at 15:52
  • $\begingroup$ You might end up getting an overfit, or none at all. Oh, and 2 objects just don't make a cluster. Just a pair. $\endgroup$ – Anony-Mousse Mar 8 '12 at 17:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.