Evaluate a cosine-measure Clustering by Davies Bouldin

User: "siamak_want"
New Altair Community Member
Updated by Jocelyn
Hi all,

I am running a process which contains a k-means operator with "cosine similarity measure". and I want to evaluate it with a "cluster distance performance" operator with "Davies Bouldin" measure. As far as I know, the Davies bouldin measure is calculated by "Eucledian distance" measure, But my cluster model has been built upon a "cosine similarity measure". SO how can I evaluate a cosine-based k-means? Any idea please?

Find more posts tagged with

Sort by:
1 - 2 of 21
    User: "siamak_want"
    New Altair Community Member
    OP
    Any explanation or idea please?
    User: "Andrew2"
    New Altair Community Member
    Hello

    I had a look at the code. I think it's this source file: CentroidBasedEvaluator.java
    private double getDaviesBouldin(CentroidClusterModel model, ExampleSet exampleSet) throws OperatorException {
    DistanceMeasure measure = new EuclideanDistance();
    I think you would have to write some Java to change how it works.

    regards

    Andrew