2

I'm trying to use Kmeans clustering, with an intent to find out clusters by weighting the attributes.

Eg. if attribute A matters less than attribute B then the output should put more weight on values of attribute B and if needed give me more clusters depending on relatively small differences in B, even if this means ignoring relatively bigger differences in values of attribute A.

Any pointers/references/examples are highly appreciated.

user2696565
  • 1,239
  • 1
  • 10
  • 14

1 Answers1

1

Since the clusters are determined using the distance function, use a distance function that adds weight to required features and conversely, reduced weights on other features.

I suggest that you can start by using a weight of 1 for the required features and ignore the rest of the features. The choice of distance function itself , such as Euclidean or Manhattan, is orthogonal to the feature weighting.

shark8me
  • 815
  • 7
  • 8