{smcl}
{hline}
help for {hi:clustervar}{right:{hi: Matthias Schonlau}}
{hline}

{title:Constructing cluster membership Variables}

{p 8 27}
{cmd:clustervar} 
{it:varlist} 
{cmd:,} 
{cmdab:al:gorithm(}{it:string}{cmd:)} 
{cmdab:dist:ance(}{it:string}{cmd:)} 
[
{cmd:min(}{it:#}{cmd:)}
{cmd:max(}{it:#}{cmd:)}
]


{title:Description}

{p}{cmd:clustervar} constructs variables named cluster{it:i} 
that indicate cluster membership. {it:i} refers to the number of clusters. 

{p}{it:varlist} contains the variables which are used as input to the cluster 
analyses.

{title:Options}

{p 0 4} {cmd:algorithm(}{it:string}{cmd:)} name of the cluster algorithm to be used.
Valid algorithms are kmeans, kmedians, singlelinkage, averagelinkage and completelinkage.
Because of ties it is possible that for a particular data set clusters with exactly i clusters
cannot be formed. Adding a very small number to all observations will usually break the tie. 
Another option is to use the min() and max() options to avoid the number
of clusters for which this happens.

{p 0 4}{cmd:distance(}{it:string}{cmd:)} specifies the distance criterion to be 
used by the cluster algorithm. Typically this is either L2 or L1.

{p 0 4} {cmd:min(}{it:#}{cmd:)} optionally specifies the smallest number of clusters.  By default this is 
set to 1. 

{p 0 4} {cmd:max(}{it:#}{cmd:)} optionally species the largest number of clusters. By default this is set to 10.

{title:Examples}

{p 4 8}{inp:. clustervar var1-var7, algorithm(kmeans) distance(L2)}

{p 4 8}{inp:. clustervar sepallen-petalwid, algorithm(averagelinkage) distance(L1) min(5) max(8)}


{title:Author}

	Matthias Schonlau, University of Waterloo
	schonlau at uwaterloo dot ca


{title:Also see}

{p 0 19}Manual:  {hi:[R] cluster}, {hi:[R] cluster dendrogram}, {hi:[R] cluster kmeans}, 
{p_end}
{p 0 19}On-line:  help for {help cluster}{p_end}