10.01.2015 Views

Package 'fpc' - open source solution for an Internet free, intelligent ...

Package 'fpc' - open source solution for an Internet free, intelligent ...

Package 'fpc' - open source solution for an Internet free, intelligent ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

18 classifdist<br />

Details<br />

classifdist is <strong>for</strong> data given as dissimilarity matrix, classifnp is <strong>for</strong> data given as n times p data<br />

matrix. The following methods are supported:<br />

"centroid" assigns observations to the cluster with closest cluster centroid as specified in argument<br />

centroids (this is associated to k-me<strong>an</strong>s <strong>an</strong>d pam/clara-clustering).<br />

"averagedist" assigns to the cluster to which <strong>an</strong> observation has the minimum average dissimilarity<br />

to all points in the cluster (this is associated to average linkage clustering).<br />

"qda" only in classifnp. Classifies by quadratic discrimin<strong>an</strong>t <strong>an</strong>alysis (this is associated to Gaussi<strong>an</strong><br />

clusters with flexible covari<strong>an</strong>ce matrices), calling qda with default settings. If qda gives<br />

<strong>an</strong> error (usually because a class was too small), lda is used.<br />

"knn" classifies by nnk nearest neighbours (<strong>for</strong> nnk=1, this is associated with single linkage clustering).<br />

Calls knn in classifnp.<br />

Value<br />

An integer vector giving cluster numbers <strong>for</strong> all observations; those <strong>for</strong> the observations already<br />

clustered in the input are the same as in the input.<br />

Author(s)<br />

Christi<strong>an</strong> Hennig http://www.homepages.ucl.ac.uk/~ucakche/<br />

See Also<br />

prediction.strength, nselectboot<br />

Examples<br />

set.seed(20000)<br />

x1

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!