mostProbableClusters: Most Probable Clusters

In cleanzr/dblinkR: R interface for dblink

Description

Computes the most probable cluster for each record in the data set based on posterior samples. The collection of most probable clusters is not guaranteed to obey transitivity of closure. To obtain a transitive clustering, one can apply the sharedMostProbableClusters function to the output of this function. In (Steorts et al. 2016), the most probable clusters are referred to as most probable maximal matching sets.

Usage

mostProbableClusters(x, ...)

## S4 method for signature 'linkagechain'
mostProbableClusters(x, ...)

## S4 method for signature 'dblinkresult'
mostProbableClusters(x, ...)

Arguments

x	a dblinkresult object as returned by runInference, or a linkagestructure as returned by loadLinkageChain.

Value

A mostprobableclusters object.

References

Steorts, R. C., Hall, R. & Fienberg, S. E. A Bayesian Approach to Graphical Record Linkage and Deduplication. JASA 111, 1660–1672 (2016).

See Also

The sharedMostProbableClusters function computes a point estimate from the most probable clusters (the output of this function), which obeys transitivity constraints.

cleanzr/dblinkR documentation built on June 13, 2021, 4:17 a.m.