clean.graph {multivariance} | R Documentation |
Given a dependence structure graph: vertices representing the multivariances of only two vertices can be turned into an edge labeled with the label of the vertex. Moreover, only subsets of the graph can be selected.
clean.graph(g, only.level = NULL, simplify.pairs = TRUE, drop.label.pairs = FALSE)
g |
graph, created by |
only.level |
integer vector, if provided all edges and dependency nodes corresponding to dependence orders not given in 'only.level' are removed |
simplify.pairs |
boolean, if true dependency nodes which are only connected to two variables are turned into edges |
drop.label.pairs |
boolean, if true the labels for edges indicating pairwise dependence are removed |
Note: The option 'only.level' works only properly for a full dependence structure graph, in the case of a clustered dependence structure graph dependency nodes representing a cluster might be removed.
graph
N = 200 y = coins(N,2) x = cbind(y,y,y) ds = dependence.structure(x,structure.type = "clustered") plot(clean.graph(ds$graph)) plot(clean.graph(ds$graph,only.level = 2)) plot(clean.graph(ds$graph,only.level = 3)) # of limited use for a clustered graph, # i.e., here the three-dependence node without edges indicates that # all edges were connected to clusters ds = dependence.structure(x,structure.type = "full") plot(clean.graph(ds$graph)) plot(clean.graph(ds$graph,drop.label.pairs = TRUE)) plot(clean.graph(ds$graph,only.level = 2)) plot(clean.graph(ds$graph,only.level = 2,drop.label.pairs = TRUE)) plot(clean.graph(ds$graph,only.level = 3))