the parameter sitting for K-mean in auto model for

Ammar_AlmasriAmmar_Almasri Member Posts: 2 Newbie
edited February 2020 in Help
when i do experimental on data set using K-means the cluster size never changes and center point as well. why they didnt change if they select ceroid randomly 
Jasmine_

Answers

  • Telcontar120Telcontar120 Moderator, RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 1,630 Unicorn
    It completely depends on the dataset characteristics.  Some datasets are "stable" in the sense that there is one clear solution for a k-means of a given size k, which will always converge to that solution.  Others are not, in which case rerunning the clustering algorithm can lead to different results.  
    You can try setting the local random seed to different numbers to see whether that makes a difference.  
    Brian T.
    Lindon Ventures 
    Data Science Consulting from Certified RapidMiner Experts
    sgenzerJasmine_
  • Ammar_AlmasriAmmar_Almasri Member Posts: 2 Newbie
    Question about when I use default sitting the K-means++ will select a good start point for that cluster never changed
    Jasmine_
Sign In or Register to comment.