I am trying to perform a cluster analysis on 50k*3 data in R. Clustering works well with hclust.vector() but I can't find a method to determine k not requiring the estimation of a distance matrix (which will be too large for R to handle in my case). Any suggestions?
Amir Hossein Poorjam
One possible option is to consider the number of clusters as an unknown parameter and estimate it along with other parametrs of your model.
Another option is to use a Bayesian nonparametric approach for clustering. This way, you let the data to choose the best number of clusters.
1 votes
1 thanks
Matthias Glatz
Plotting the residual variance against the number of clusters usually gives good answers.
1 votes
1 thanks
- Badges
- Science topic
- Similar topics
- Mathematics
- Algebra
- Matrix
More Tom Broekel's questions See All
Similar questions and discussions