refine: Refining a Classification by Re-Assigning Memberships
In optpart: Optimal Partitioning of Similarity Relations

Description Usage Arguments Value Note Author(s) References Examples

Refine allows you to re-assign specific elements of a classification from one class or cluster to another. In the default case, you simply interactively enter sample IDs and give a new cluster assignment. For PCO and NMDS ordinations, you do the assignments with a mouse.

## Default S3 method:
refine(comm,clustering,...)
## S3 method for class 'dsvord'
refine(x,clustering,ax=1,ay=2,...)

`comm`	a community data.frame
`x`	an ordination of class ‘dsvord’
`clustering`	a clustering identity or membership vector
`ax`	the X axis of the ordination
`ay`	the Y axis of the ordination
`...`	ancillary arguments to allow differing numbers of arguments

a list object of class ‘clustering’ with one component.

clustering

a numeric vector giving the cluster assignment for each sample

There are many, many ways to produce classifications in R, including several in package ‘optpart’. refine is designed to take one of these classifications and polish it by making relatively few transfers. The ordination-based routines allow visual assessment of cluster validity, although in reduced dimensionality, which can be misleading.

David W. Roberts droberts@montana.edu

http://ecology.msu.montana.edu/labdsv/R

## Not run: data(shoshveg)
## Not run: dis.bc <- dsvdis(shoshveg,'bray')
## Not run: opt.5 <- optpart(5,dis.bc)
## Not run: nmds.bc <- nmds(dis.bc)
## Not run: plot(nmds.bc)
## Not run: res <- refine(nmds.bc,opt.5)