swh:1:snp:0f920b1e114986636ba2e45b5c1a83473fb6cf12
Tip revision: 04f271f14fba8652b1ba48e065b58dbbc8b36bbd authored by Piotr Romanski on 11 April 2009, 00:00:00 UTC
version 0.15
version 0.15
Tip revision: 04f271f
correlation.Rd
\name{correlation}
\alias{linear.correlation}
\alias{rank.correlation}
\title{ Correlation filter}
\description{
The algorithm finds weights of continous attributes basing on their correlation with continous class attribute.
}
\usage{
linear.correlation(formula, data)
rank.correlation(formula, data)
}
\arguments{
\item{formula}{ a symbolic description of a model }
\item{data}{ data to process }
}
\details{
\code{linear.correlation} uses Pearson's correlation
\code{rank.correlation} uses Spearman's correlation
Rows with \code{NA} values are not taken into consideration.
}
\value{
a data.frame containing the worth of attributes in the first column and their names as row names
}
\author{ Piotr Romanski }
\examples{
library(mlbench)
data(BostonHousing)
d=BostonHousing[-4] # only numeric variables
weights <- linear.correlation(medv~., d)
print(weights)
subset <- cutoff.k(weights, 3)
f <- as.simple.formula(subset, "medv")
print(f)
weights <- rank.correlation(medv~., d)
print(weights)
subset <- cutoff.k(weights, 3)
f <- as.simple.formula(subset, "medv")
print(f)
}