1
$\begingroup$

I've been playing around with some data, and I've created a decently large similarity matrix (each column value for a row represents how similar the document corresponding to the column is to the document corresponding to the row) and I'm trying to scale this data down to 2 dimensions so I can visualize it on a scatter plot, where each point represents a document, and the closer two documents are, the closer they would be on the scatter plot. I've tried using Sammon Mapping to accomplish this, but the problem is it takes a significant amount of time to run. Any ideas?

$\endgroup$

1 Answer 1

1
$\begingroup$

You may use a sampling based method, in which you get an approximation of the actual result by combining the partial results for samples extracted from your data. There are several methods based on sampling, but you could take a look to the following link as an example, so you can understand what I am talking about:

A Fast Approximation to Multidimensional Scaling

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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