network: Measures of network

In ECoL: Complexity Measures for Supervised Problems

Description

Classification task. The network measures represent the dataset as a graph and extract structural information from it. The transformation between raw data and the graph representation is based on the epsilon-NN algorithm. Next, a post-processing step is applied to the graph, pruning edges between examples of opposite classes.

Usage

network(...)

## Default S3 method:
network(x, y, measures = "all", eps = 0.15,
  summary = c("mean", "sd"), ...)

## S3 method for class 'formula'
network(formula, data, measures = "all", eps = 0.15,
  summary = c("mean", "sd"), ...)

Arguments

...	Not used.
x	A data.frame contained only the input attributes.
y	A factor response vector with one label for each row/component of x.
measures	A list of measures names or "all" to include all them.
eps	The percentage of nodes in the graph to be connected.
summary	A list of summarization functions or empty for all values. See summarization method to more information. (Default: c("mean", "sd"))
formula	A formula to define the class column.
data	A data.frame dataset contained the input attributes and class.

Details

The following measures are allowed for this method:

"Density"

Average Density of the network (Density) represents the number of edges in the graph, divided by the maximum number of edges between pairs of data points.

"ClsCoef"

Clustering coefficient (ClsCoef) averages the clustering tendency of the vertexes by the ratio of existent edges between its neighbors and the total number of edges that could possibly exist between them.

"Hubs"

Hubs score (Hubs) is given by the number of connections it has to other nodes, weighted by the number of connections these neighbors have.

Value

A list named by the requested network measure.

References

Gleison Morais and Ronaldo C Prati. (2013). Complex Network Measures for Data Set Characterization. In 2nd Brazilian Conference on Intelligent Systems (BRACIS). 12–18.

Luis P F Garcia, Andre C P L F de Carvalho and Ana C Lorena. (2015). Effect of label noise in the complexity of classification problems. Neurocomputing 160, 108–119.

See Also

Other complexity-measures: balance, correlation, dimensionality, linearity, neighborhood, overlapping, smoothness

Examples

## Extract all network measures for classification task
data(iris)
network(Species ~ ., iris)

Example output

$Density
[1] 0.8332886

$ClsCoef
[1] 0.267977

$Hubs
     mean        sd 
0.8380508 0.2753319 

ECoL documentation built on Nov. 5, 2019, 9:07 a.m.