Options

Simple process Q: Clustering with weights

villepohjanheimvillepohjanheim Member Posts: 2 Contributor I
edited July 2019 in Help
Hi all

A real rookie question here. I tried searching the forum, but maybe i just don't know the right words. So please bear with..

I'm attempting to do a simple k-means clustering in RM (4.5) with answers/rows weighted. As the operator doesn't provide such option, how should I go about this?

Thanks for all answers/links/suggestions

Answers

  • Options
    landland RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 2,531 Unicorn
    Hi,
    currently KMeans does not support weighted examples. The only way to go is to insert each example multiple times. If an example is twice as often part of the example set as another, it has the doubled weight on the distance calculation. Of course this way is hmmm not really desireable, because the runtime will increase with number of examples...

    For a great number of examples or arbitrary weights, the only real way to go would be to extend the operator. Either write it on your own, or get it done by us for little money.

    Greetings,
      Sebastian
Sign In or Register to comment.