Skip to contents

Expected Influence (one-step)

Source: R/centrality.R
centrality_expected_influence_1.Rd

Signed-weight sum of a node's edges (Robinaugh, Millner & McNally 2016). The appropriate centrality for networks with positive and negative edges (partial-correlation, glasso, signed correlation networks) where strength — which takes absolute values — can be misleading.

Usage

centrality_expected_influence_1(x, mode = "out", ...)

Arguments

x

Network input (matrix, igraph, network, cograph_network, tna object).

mode

One of "all", "in", "out" for directed graphs. Default "out".

...

Additional arguments passed to centrality.

Value

Named numeric vector of expected-influence values (signed).

References

Robinaugh DJ, Millner AJ, McNally RJ (2016). Identifying highly influential nodes in the complicated grief network. Journal of Abnormal Psychology, 125(6), 747-757.

See also

centrality_expected_influence_2 for the two-step variant, centrality_strength for the unsigned analogue.

Examples

# Signed weight matrix (partial correlations, for example)
W <- matrix(c( 0.0,  0.5, -0.3,  0.2,
               0.5,  0.0,  0.4, -0.1,
              -0.3,  0.4,  0.0,  0.6,
               0.2, -0.1,  0.6,  0.0), 4, 4, byrow = TRUE)
rownames(W) <- colnames(W) <- c("A", "B", "C", "D")
centrality_expected_influence_1(W)
#>   A   B   C   D 
#> 0.4 0.8 0.7 0.7