-1
$\begingroup$

I'm planning on extracting a number of word vector distances from a data set, and I want to be able to detect clusters within that set, with an undefined number of clusters that are dynamically defined based on a distance variable.

In general terms, what are my options I can look into?

$\endgroup$
1
$\begingroup$

You can try k-means algorithm. All you need to tune there is the distance function and the number of clusters. It is pretty simple to understand too.

$\endgroup$
  • $\begingroup$ What would be a good way to automatically extract clusters that fall within a certain distance without manually defining? If there are an unknown number of clusters $\endgroup$ – jmhead Jul 12 '18 at 4:48
  • $\begingroup$ You may need to start a "for loop" on the number of clusters and visualize the mean of distances of each item to the assigned clusters. Then you will find the point where it makes no sense to increase the number of clusters as the "error" does not decrease much. $\endgroup$ – Timur Jul 13 '18 at 5:35

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