CNN: Condensed Nearest Neighbors

Description Usage Arguments Details Value References See Also Examples

View source: R/CNN.R

Description

Similarity-based method designed to select the most relevant instances for subsequent classification with a nearest neighbor rule. For more information, see 'Details' and 'References' sections.

Usage

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

## Default S3 method:
CNN(x, 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.

classColumn

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

Details

CNN searches for a 'consistent subset' of the provided dataset, i.e. a subset that is enough for correctly classifying the rest of instances by means of 1-NN. To do so, CNN stores the first instance and goes for a first sweep over the dataset, adding to the stored bag those instances which are not correctly classified by 1-NN taking the stored bag as training set. Then, the process is iterated until all non-stored instances are correctly classified.

Although CNN is not strictly a label noise filter, it is included here for completeness, since the origins of noise filters are connected with instance selection algorithms.

Value

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

References

Hart P. (May, 1968): The condensed nearest neighbor rule, IEEE Trans. Inf. Theory, vol. 14, no. 5, pp. 515-516.

See Also

RNN

Examples

1
2
3
4
5
6
7
8
9
# Next example is not run in order to save time
## Not run: 
data(iris)
out <- CNN(iris)
print(out)
length(out$remIdx)
identical(out$cleanData, iris[setdiff(1:nrow(iris),out$remIdx),])

## End(Not run)

Example output

Call:
CNN(x = iris)

Results:
Number of removed instances: 131 (87.33333 %)
Number of repaired instances: 0 (0 %)
[1] 131
[1] TRUE

NoiseFiltersR documentation built on May 29, 2017, 3:12 p.m.