https://github.com/cran/RecordLinkage
Tip revision: 91236cabdeb855d9c01a2714a751518c47acdc6b authored by Andreas Borg on 04 March 2019, 14:20:44 UTC
version 0.4-10.2
version 0.4-10.2
Tip revision: 91236ca
gpdEst.Rd
\name{gpdEst}
\Rdversion{1.1}
\alias{gpdEst}
\title{
Estimate Threshold from Pareto Distribution
}
\description{Fits a Pareto distribution to the distribution of weights
and calculates a quantile on the fitted model as classification threshold.
}
\usage{
gpdEst(Wdata, thresh = -Inf, quantil = 0.95)
}
\arguments{
\item{Wdata}{
A numeric vector representing weights of record pairs.
}
\item{thresh}{
Threshold for exceedances.
}
\item{quantil}{
A real number between 0 and 1. The desired quantil.
}
}
\details{
The weights that exceed \code{thresh} are fitted to a
generalized Pareto distribution (GPD). The estimated parameters \code{shape}
and \code{scale} are used to calculate a classification threshold by the
formula
\deqn{\mathit{thresh}+\frac{\mathit{scale}}{\mathit{shape}}
((\frac{n}{k}(1-\mathit{quantil}))^{-\mathit{shape}} -1)}{
thresh+scale/shape*((n/k*(1-quantil))^(-shape) -1)}
where \eqn{n} is the total number of weights and \eqn{k} the number of
exceedances.
}
\value{
A real number representing the resulting classification threshold. It is
assured that the threshold lies in a reasonable range.
}
\author{
Murat Sariyar
}
\seealso{
\code{\link{getParetoThreshold}} for user-level function
}
\keyword{models}
\keyword{misc}