ldstatsHD (version 1.0.1)

pcLambdaSelection: Path Connectivity regularization parameter selection

Description

pcLambdaSelection is a function designed to select the regularization parameter in graphical models. It selects the graph which captures the biggest drop in graph connectivity.

Usage

pcLambdaSelection(obj)

Arguments

obj

an object of class huge, camel.tiger or wfgl.

Value

An object of class lambdaSelection containing the following components:

opt.lambda

optimal lambda.

crit.coef

coefficients for each lambda given the criterion PC.

criterion

with value "PC".

Details

Path Connectivity (PC) algorithm finds \(\lambda\) by maximizing the biggest drop of connectivity in estimated graphs. We define connectivity by the average geodesic distance between pairs of nodes (see graphDist).

PC gives a fast and suitable way to select \(\lambda\) when there are distinct clusters in the data. Given two graphs, corresponding to two consecutive \(\lambda\)'s, the difference between the average geodesic distance will be large if the first graph contains edges that connect different clusters which are not present in the second graph.

Note that PC should be used when fitting graphical models with an equidistant sequence for \(\lambda\).

References

Caballe, A., N. Bochkina, and C. Mayer (2016). Selection of the Regularization Parameter in Graphical Models using network charactaristics. eprint arXiv:1509.05326, 1-25.

See Also

lambdaSelection for other lambda selection approaches.

Examples

# NOT RUN {
# example to use pc function
EX1         <- pcorSimulator(nobs = 70, nclusters = 3, nnodesxcluster = c(40,30,20), 
                             pattern = "powerLaw")
y           <- EX1$y
Lambda.SEQ  <- seq(.25,0.70,length.out = 40)
out3        <- huge(y, method = "mb", lambda = Lambda.SEQ)
PC.COEF     <- pcLambdaSelection(out3)
print(PC.COEF)
 
 
# }