conf.edge.lower: Lower limit of the confidence of an edge

In MXM: Feature Selection (Including Multiple Solutions) and Bayesian Networks

Lower limit of the confidence of an edge

Description

Lower limit of the confidence of an edge.

Usage

conf.edge.lower(p)

Arguments

p	A numerical vector with the proportion of times an edge was found in the bootstrapped PC algorithm or the confidence of the edge returned by bn.skel.utils2.

Details

After having performed PC algorithm many times in the bootstrap samples (using pc.skel.boot for example) you get a symmetric matrix with the proportion of times an edge was discovered. Take the lower (or upper) triangular elements of that matrix and pass them as input in this function. This will tell you the minimum proportion required to be confident that an edge is trully significant.

Value

The estimated cutoff limit above which an edge can be deemed significant.

Author(s)

Michail Tsagris

R implementation and documentation: Michail Tsagris [e-mail]

References

Scutari M. and Nagarajan R. (2013). Identifying significant edges in graphical models of molecular networks. Artifficial Intelligence in Medicine, 57: 207-217.

See Also

pc.skel.boot

Examples

y <- rdag2(200, p = 40, nei = 3)
x <- y$x
g <- pc.skel.boot(x, R = 199)$Gboot
a <- g[ lower.tri(g) ]
conf.edge.lower(a)

MXM documentation built on Aug. 25, 2022, 9:05 a.m.