cograph (version 2.7.2)

select_nodes: Select Nodes with Lazy Centrality Computation

Description

A more nuanced node selection function that improves upon filter_nodes() with lazy centrality computation (only computes measures actually referenced), multiple selection modes, and global context variables for structural awareness.

Usage

select_nodes(
  x,
  ...,
  name = NULL,
  index = NULL,
  top = NULL,
  by = "degree",
  neighbors_of = NULL,
  order = 1L,
  component = NULL,
  keep_edges = c("internal", "none"),
  keep_format = FALSE,
  directed = NULL,
  .keep_edges = NULL
)

Value

A cograph_network object with selected nodes. If keep_format = TRUE, matrix, igraph, and statnet network inputs are converted back to that type.

Arguments

x

Network input: cograph_network, matrix, igraph, network, or tna object.

...

Filter expressions using node columns, centrality measures, or global context variables. Centrality measures are computed lazily (only those actually referenced). Available variables:

Node columns

All columns in the nodes dataframe: id, label, name, x, y, inits, color, plus any custom

Centrality measures

degree, indegree, outdegree, strength, instrength, outstrength, betweenness, closeness, eigenvector, pagerank, hub, authority, coreness. Any other measure centrality() computes can be named too; see list_centralities().

Global context

component, component_size, is_largest_component, neighborhood_size, k_core, is_articulation, is_bridge_endpoint

Predicates

is_isolated, is_source, is_sink, is_leaf, is_cut, local_transitivity, local_triangles

name

Character vector. Select nodes by name/label.

index

Integer vector. Select nodes by index (1-based).

top

Integer. Select top N nodes by centrality measure.

by

Character. Centrality measure for top selection. Default "degree".

neighbors_of

Character or integer. Select neighbors of these nodes (by name or index).

order

Integer. Neighborhood order (1 = direct neighbors, 2 = neighbors of neighbors, etc.). Default 1.

component

Selection mode for connected components:

"largest"

Select nodes in the largest connected component

Integer

Select nodes in component with this ID

Character

Select component containing node with this name

keep_edges

How to handle edges. One of:

"internal"

(default) Keep only edges between remaining nodes

"none"

Remove all edges

keep_format

Logical. If TRUE, matrix, igraph, and statnet network inputs are returned in that format. Default FALSE returns cograph_network.

directed

Logical or NULL. If NULL (default), auto-detect.

.keep_edges

Deprecated. Use keep_edges.

Details

Selection modes are combined with AND logic (like tidygraph/dplyr):

• select_nodes(x, top = 10, component = "largest") selects top 10 nodes within the largest component

• All criteria must be satisfied for a node to be selected

Centrality measures are computed lazily - only measures actually referenced in expressions or the by parameter are computed. This makes select_nodes() faster than filter_nodes() for large networks.

For networks with negative edge weights, betweenness, closeness and pagerank are undefined and return NA, with a cograph_negative_weights warning.

See Also

filter_nodes, select_neighbors, select_component, select_top

Examples

adj <- matrix(c(0, .5, .8, 0, .5, 0, .3, .6,
                .8, .3, 0, .4, 0, .6, .4, 0), 4, 4, byrow = TRUE)
rownames(adj) <- colnames(adj) <- c("A", "B", "C", "D")

select_nodes(adj, degree >= 3)
select_nodes(adj, top = 2, by = "pagerank")
select_nodes(adj, neighbors_of = "A", order = 2)
select_nodes(adj, component = "largest")