GE: Generalized Edition

Description

Similarity-based filter for removing or repairing label noise from a dataset as a preprocessing step of classification. For more information, see 'Details' and 'References' sections.

Usage

1
2
3
4
5
6
## S3 method for class 'formula'
GE(formula, data, ...)

## Default S3 method:
GE(x, k = 5, kk = ceiling(k/2), classColumn = ncol(x),
  ...)

Arguments

formula

A formula describing the classification variable and the attributes to be used.

data, x

Data frame containing the tranining dataset to be filtered.

...

Optional parameters to be passed to other methods.

k

Number of nearest neighbors to be considered.

kk

Minimum size for local majority class in order to relabel an instance.

classColumn

positive integer indicating the column which contains the (factor of) classes. By default, the last column is considered.

Details

GE is a generalization of ENN that integrates the possibility of 'repairing' or 'relabeling' instances rather than only 'removing'. For each instance, GE considers its k-1 neighbors and the instance itself. If there are at least kk examples from the same class, the instance is relabeled with that class (which could be its own). Otherwise, it is removed.

Value

An object of class filter, which is a list with seven components:

References

Koplowitz J., Brown T. A. (1981): On the relation of performance to editing in nearest neighbor rules. Pattern Recognition, 13(3), 251-255.

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
# Next example is not run in order to save time
## Not run: 
data(iris)
out <- GE(iris)
summary(out, explicit = TRUE)
# We check that the process was correct
irisCopy <- iris
irisCopy[out$repIdx,5] <- out$repLab
cleanData <- irisCopy[setdiff(1:nrow(iris),out$remIdx),]
identical(out$cleanData,cleanData)

## End(Not run)

Questions? Problems? Suggestions? or email at ian@mutexlabs.com.

All documentation is copyright its authors; we didn't write any of that.