-1

I have implement 3 clustering algorithms in R (PAM, k-means and hierarchical). I want to find which parameters produce the best results of each algorithm.

I have no idea how to do it in R.

Does anyone know how to do it?

Thank you for your help.

Paolina
  • 3
  • 4
  • Define "best results" as it applies to clustering – Andrie May 09 '16 at 10:05
  • For instance, I have 7 clusters and I need to compare it with the class variable. How to compare whether the clustering is applied effectively on the the algorithm. Is there any tool that the R offers or any percentage illustrate the accuracy? – Paolina May 09 '16 at 10:08
  • check out NBclust! – Chirayu Chamoli May 09 '16 at 10:16
  • The number of clusters used was predetermined (7) what I am looking for is how to compare the validity of the methods. – Paolina May 09 '16 at 10:21

1 Answers1

-1

You need a statistic which can be calculated for all 3 Clustering methods like BIC criterion.

The model with the lowest BIC would be best. However this criterion should be used with caution, as overfitting is a frequent problem.

Otto_K
  • 347
  • 2
  • 6