I'm using agnes to cluster my data using the ward method. The dendogram returns clusters ordered by height and an agglomerative coefficient. My reading of Struyf, Hubert, & Rousseeuw (1997) is that agnes will provide a sense of the overall structure, but the silhouette coefficient in pam will give a heuristic on the number of clusters. Thoughts?
We developed an R package called 'clValid' which incorporates a variety of validation approaches, including internal (silhouette width, connectivity, Dunn index) and stability measures (based on leave one out cross-validation). The function can directly incorporate a number of clustering methods, including agnes. We published a paper describing the approach in the Journal of Statistical Software (2008, 25:4) which is freely available. There are also other packages on CRAN which deal with cluster validation, including 'clv' and 'clusteval', among others.
We developed an R package called 'clValid' which incorporates a variety of validation approaches, including internal (silhouette width, connectivity, Dunn index) and stability measures (based on leave one out cross-validation). The function can directly incorporate a number of clustering methods, including agnes. We published a paper describing the approach in the Journal of Statistical Software (2008, 25:4) which is freely available. There are also other packages on CRAN which deal with cluster validation, including 'clv' and 'clusteval', among others.
Thanks! So far so good. You have some of the best documentation I've seen in the R universe.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Software
- R Statistical Package