Evaluation of Support Vector Clustering

Muhammed_Fatih_Muhammed_Fatih_ Member Posts: 93 Maven
edited June 2020 in Help
Hello Community, 

is there a possibility within RapidMiner to (iteratively) evaluate the parameters of Support Vector Clustering (SVC)? 

Thank you in advance for your responses! 

Best regards!  

Answers

  • hbajpaihbajpai Member Posts: 102 Unicorn
    Muhammed_Fatih_ You can use Optimize Parameters operator to evaluate parameters at different levels. Just in case you find a parameter that is not in default list, you can use macro with list to initialize and iterate over them. 
    Best,
    Harshit
  • Muhammed_Fatih_Muhammed_Fatih_ Member Posts: 93 Maven
    Hello @hbajpai

    thank you for your answer! I will try it out. In this connection, do you know a performance measure which could be evaluated with SVC? 

    Thank you in advance for your feedback! 

    Best regards!
  • hbajpaihbajpai Member Posts: 102 Unicorn
    Hey @Muhammed_Fatih_

    The evaluation of clustering is always a tricky aspect. In most cases, it depends on the problem at hand, the hypothesis behind the clustering and whether we have ground truth available. Having said that, you can check out the performance operators in the segmentation section of RM and see if on of those works for you. 


    Best,
    Harshit
  • Muhammed_Fatih_Muhammed_Fatih_ Member Posts: 93 Maven
    Hi @hbajpai

    Thank you for the hint. Unfortunately, I already looked up in the segmentation section of RM regarding performance operators but I couldn't find an appropriate one especially for Support Vector Clusters. Most of them target centroid based Clustering approaches like e.g. kMeans. 

    Did you or rather the community use one of them for Support Vector Clustering before? 

    Thank you in advance for your feedback! 

    Best regards!
Sign In or Register to comment.