Skip to main content
Cross Validated

Return to Question

expand abbrev on first use
Link
Glen_b
Glen_b
  • 290.5k
  • 37
  • 652
  • 1.1k

PAM Partitioning Around Medoids (PAM) with Gower distance matrix

Tweeted twitter.com/#!/StackStats/status/555177767055425537
capitalization, punctuation, extra tag
Source Link
Nick Stauner
Nick Stauner
  • 12.4k
  • 5
  • 53
  • 110

My data is is mostly continuous but has one binary variable. I tried the pam algorithm in R with the Gower index, but the number of clusters that give the best silhouette width is 2 -– allowing the binary variable to completely dominate the result.

  • Is PAM the wrong approach?
  • Is it okOK to choose a higher k just because it will give more meaningful results?

My data is is mostly continuous but has one binary variable. I tried the pam algorithm in R with the Gower index but the number of clusters that give the best silhouette width is 2 - allowing the binary variable to completely dominate the result.

  • Is PAM the wrong approach?
  • Is it ok to choose a higher k just because it will give more meaningful results?

My data is is mostly continuous but has one binary variable. I tried the pam algorithm in R with the Gower index, but the number of clusters that give the best silhouette width is 2 – allowing the binary variable to completely dominate the result.

  • Is PAM the wrong approach?
  • Is it OK to choose a higher k just because it will give more meaningful results?
Formatting
Source Link
edit approved Jan 25, 2014 at 17:59
user35349
edit approved Jan 25, 2014 at 17:59
user35349

My data is is mostly continuous but has one binary variable,. I tried the pampam algorithm in R with the Gower index but the number of clusters that give the best silhouette width is 2 - allowing the binary variable to completely dominate the result. Is PAM the wrong approach? Is it ok to choose a higher k just because it will give more meaningful results?

  • Is PAM the wrong approach?
  • Is it ok to choose a higher k just because it will give more meaningful results?

My data is is mostly continuous but has one binary variable, I tried the pam algorithm in R with the Gower index but the number of clusters that give the best silhouette width is 2 - allowing the binary variable to completely dominate the result. Is PAM the wrong approach? Is it ok to choose a higher k just because it will give more meaningful results?

My data is is mostly continuous but has one binary variable. I tried the pam algorithm in R with the Gower index but the number of clusters that give the best silhouette width is 2 - allowing the binary variable to completely dominate the result.

  • Is PAM the wrong approach?
  • Is it ok to choose a higher k just because it will give more meaningful results?
Source Link
Jin
Jin
  • 61
  • 2