classifyUnsup {RecordLinkage}

Unsupervised Classification

Description

Classify record pairs with unsupervised clustering methods.

Usage

classifyUnsup(rpairs, method, ...)

Arguments

rpairs

Object of type RecLinkData. The data to classify.

method

The classification method to use. One of "kmeans", "bclust".

...

Further arguments for the classification method

Details

A clustering algorithm is applied to find clusters in the comparison patterns. In the case of two clusters (the default), the cluster further from the origin (i.e. representing higher similarity values) is interpreted as the set of links, the other as the set of non-links.

Supported methods are:

kmeans

K-means clustering, see kmeans.

bclust

Bagged clustering, see bclust.

Value

An object of class "RecLinkResult" that represents a copy of newdata with element rpairs$prediction, which stores the classification result, as addendum.

Author(s)

Andreas Borg, Murat Sariyar

See Also

trainSupv and classifySupv for supervised classification.

Examples

# Classification with bclust
data(RLdata500)
rpairs=compare.dedup(RLdata500, identity=identity.RLdata500,
                    blockfld=list(1,3,5,6,7))
result=classifyUnsup(rpairs,method="bclust")
summary(result)