aggregate-graphs: Aggregate a list of causality.graphs into a single object
In tzimiskes/causality: Examine The Causal Structure of Data

Description Usage Arguments Details Author(s) Examples

aggregate_graphs will combine graphs into an "aggregated.causality.graph" object which can be used for bootstrapping and jackknifing causal discovery algorithms among other things.

1
2
3

aggregate_graphs(graphs, filter = 0.1, weights = NULL)

coalesce(aggregated.graph)

`graphs`	A list of causality.graphs. Each graph must have the same nodes.
`filter`	Numeric between 0 and 1. aggregate_graphs filters out rows that sum to less than filter.
`weights`	An optional vector of graph weights to be used in the aggregation process. If weights is not supplied, it defaults to `length(graphs)`. If weights is non NULL, the length of weights must equal the length of graphs, the sum of weights must be greater than 0, and each weight must be non-negative.
`aggregated.graph`	An aggregated.causality.graph that you wish to turn into a causality graph.

TODO

Alexander Rix

# jackknife ges
n.jks   <- 10
jk.frac <- .9
n.obs   <- nrow(ecoli.df)
graphs  <- vector("list", n.jks)
for (i in 1:n.jks) {
 ecoli.jk <- ecoli.df[sample(n.obs, n.obs * jk.frac, replace = FALSE),]
 graphs[[i]] <- ges(ecoli.jk, "bic", penalty = 1)
}
aggregated <- aggregate_graphs(graphs)