goodness of fit for agglomerative hierarchical clustering

We generally do not know the number of classes in advance. Thus, we need to test for goodness of fit or we may just be looking at statistical noise. Please consider adding the cophenetic correlation coefficient (Sokal and Rohlf, 1962 or Farris) or some other measure of goodness of fit.

  • Guest
  • Sep 8 2016
  • Shipped
  • Attach files
  • Ricardo Araujo commented
    November 02, 2016 13:21

    very good idea

  • Adriano CRUZ commented
    January 02, 2017 13:30

    It is very good idea I agree

  • Jean Paul Maalouf commented
    March 29, 2017 12:02

    The cophenetic correlation matrix and coefficient have been added in version 2017.2