corrDependency: Linear correlation-based dependency measure

In janusza/RmRMR: Feature Subset Selection Using the mRMR Approach

Description

This is an exemplary function for measuring inter-attribute dependency and dependency between attributes and the dacisions. It is used by default by the provided implementation of the mRMR framework.

Usage

corrDependency(x, y, ...)

Arguments

x, y	two numeric attributes or an attribute and a numeric decision (binary and integer valued decisions are also accepted).
...	optional arguments to the cor function.

Value

a numeric value expressing linear dependency between x and y

Author(s)

Andrzej Janusz

References

Mark Hall. Correlation-based Feature Selection for Machine Learning. PhD thesis, University of Waikato, 1999.

Examples

#############################################
data(methaneSampleData)

## an experiment on a sample from the data used in a data mining competition - 
## IJCRS'15 Data Challenge: Mining Data from Coal Mines 
## (https://knowledgepit.fedcsis.org/contest/view.php?id=109).
## The whole data set can be downloaded from the competition web page.

mrmrAttrs = mRMRfs(dataT = methaneData$methaneTraining,
                   target = methaneData$methaneTrainingLabels[, as.integer(V2 == 'warning')],
                   dependencyF = corrDependency)

mrmrAttrs

janusza/RmRMR documentation built on May 18, 2019, 2:39 p.m.