| centrality | R Documentation |
Computes centrality measures for nodes in a network and returns a tidy data frame. Accepts matrices, edge-list data frames, igraph objects, cograph_network, or tna objects.
centrality(
x,
type = c("basic", "extended", "all"),
measures = NULL,
include = NULL,
mode = "all",
normalized = FALSE,
weighted = TRUE,
directed = NULL,
loops = TRUE,
simplify = "sum",
digits = NULL,
sort_by = NULL,
cutoff = -1,
invert_weights = NULL,
alpha = 1,
damping = 0.85,
personalized = NULL,
transitivity_type = "local",
isolates = "nan",
lambda = 1,
diffusion_method = NULL,
k = 3,
states = NULL,
decay_parameter = 0.5,
dmnc_epsilon = 1.7,
membership = NULL,
katz_alpha = 0.1,
hubbell_weight = 0.5,
shapley_k = 2,
shapley_cutoff = 2,
s_shell_a = 0.5,
discount_p = 0.01,
ncvote_theta = 0.5,
comm_r = "max_intra",
ld_radius = 2,
enrenew_depth = 2,
voterank_lambda = 0.1,
contraction_rho = 5,
wks_alpha = 1,
wks_beta = 1,
renewed_threshold = 2,
kpath_k = 3,
kpath_len = 3,
epc_threshold = 0.5,
epc_runs = 1000,
epc_seed = NULL,
betweenness_delta = 1,
closeness_delta = 1,
gravity_mass = "kshell",
gravity_radius = 3,
mdd_lambda = 0.7,
volume_radius = 2,
diffusion_q = 1,
diffusion_steps = 3,
ds_beta = 0.1,
ds_mu = 1,
ds_steps = 5,
cda_alpha = 0.5,
icc_alpha = 0.2,
exogenous_base = "reverse_closeness",
wlr_alpha = 1,
alr_h_mode = "all",
grc_gamma = 1,
rwd_decay = 0.5,
rwd_node_weights = NULL,
linerank_aggregation = "probability",
bridging_steps = 2,
bridging_values = NULL,
proximal_variant = "source",
exf_alpha = 2,
beta_direction = "positive",
ninl_order = 3,
ninl_radius = NULL,
map_flow = "unrecorded",
map_convention = "paper",
sr_prior = 0,
mcgm_radius = 2,
mcgm_alpha = NULL,
dkgm_radius = 2,
nd_order = 2,
nd_decay = 0.2,
nd_mass = "degree",
ira_mass = "coreness",
ira_alpha = 1,
ira_tol = 1e-06,
ira_max_iter = 1000,
iira_beta = 0.2,
iira_steps = 50,
hcc_delta = 0.5,
lhc_radius = 2,
tpr_alpha = 0.85,
tpr_k = 0.85,
tpr_decay = 1,
tpr_tol = 1e-14,
tpr_max_iter = 1000,
rsp_beta = 0.01,
rsp_cost = c("inverse", "weight"),
re_indexes = c("degree", "closeness", "betweenness", "constraint"),
re_negative = NULL,
tna_network = NULL,
psych_network = NULL,
shared_nodes = FALSE,
...
)
x |
Network input (matrix, edge-list data frame, igraph, network, cograph_network, tna object). |
type |
Character scalar selecting a curated tier of measures when
Passing |
measures |
Character vector of measure names to compute. When
The remaining measures are described under Details. |
include |
Character vector of costly measures to add to a tier, or
|
mode |
For directed networks: "all" (default), "in", or "out".
Affects the mode-aware measures, whose output columns carry a mode
suffix. These include degree, strength, closeness, eccentricity,
coreness, harmonic, diffusion, leverage, kreach, the distance-based
measures, most community-aware measures, and expected influence (the
|
normalized |
Logical. If |
weighted |
Logical. Use edge weights if available. Default TRUE.
With |
directed |
Logical or NULL. If NULL (default), auto-detect from matrix symmetry. Set TRUE to force directed, FALSE to force undirected. |
loops |
Logical. If TRUE (default), keep self-loops. Set to FALSE to
remove them before calculation. |
simplify |
How to combine multiple edges between the same node pair
(possible only from edge-list, cograph_network or igraph input).
Options: "sum" (default), "mean", "max", "min". |
digits |
Integer or NULL. Round all numeric columns to this many decimal places. Default NULL (no rounding). |
sort_by |
Character or NULL. Column name to sort results by (descending order). Default NULL (original node order). A name that is not a column of the result raises an error. |
cutoff |
Maximum path length to consider for betweenness, closeness, harmonic centrality and the distance-based closeness variants (radiality, lin, decay, residual_closeness, dangalchev, generalized_closeness, harary, average_distance, barycenter, wiener, centroid, closeness_vitality, delta_closeness). Default -1 (no limit). A positive value ignores longer paths, which shortens computation on large networks and changes the values. |
invert_weights |
Logical or NULL. Whether path- and distance-based
measures (for example betweenness, closeness, harmonic, k-reach,
radiality, decay, stress, flow betweenness and related variants) invert
the weights, so that higher weights mean shorter paths. A message reports
the inversion. Eccentricity always reads the raw edge weights as
distances. The default |
alpha |
Numeric. Exponent for weight transformation when |
damping |
PageRank damping factor. Default 0.85. Must be between 0 and 1. |
personalized |
Non-negative numeric vector of reset probabilities for personalized PageRank, one value per node. A named vector is matched to the node names, an unnamed one is taken in node order. The vector is rescaled to sum to 1. Default NULL (standard PageRank). |
transitivity_type |
Type of transitivity to calculate: "local" (default),
"global", "undirected", "localundirected", "barrat" (weighted),
"weighted", or "onnela". The first six follow the conventions of
|
isolates |
Value of local transitivity at nodes where it is undefined: "nan" (default) returns NaN, "zero" returns 0. |
lambda |
Diffusion scaling factor for diffusion centrality. Default 1.
Only used when |
diffusion_method |
Character or NULL. Selects the diffusion-centrality
formula. |
k |
Distance bound for |
states |
Numeric vector of percolation states for percolation
centrality, one per node. Each value represents how "activated" or
"infected" a node is. A named vector is matched to the node labels,
missing nodes get state 1, and values are clipped to 0-1. Default NULL
gives every node state 1, which makes percolation equal to betweenness
divided by |
decay_parameter |
Numeric. Decay parameter for decay and generalized
closeness centrality, strictly between 0 and 1; smaller values discount
distant nodes more. Other values raise a |
dmnc_epsilon |
Numeric. Epsilon exponent for DMNC (Density of Maximum
Neighborhood Component), a single positive number. Default 1.7, as
recommended by Lin et al. (2008). centiserve uses 1.67. Other values
raise a |
membership |
Integer vector of community assignments (one per node) for
the community-aware measures listed under |
katz_alpha |
Attenuation factor for Katz centrality. The Katz series
converges only for |
hubbell_weight |
Weight factor |
shapley_k |
Neighbor threshold |
shapley_cutoff |
Hop cutoff for |
s_shell_a |
Exponent of the asymmetric link weights for
|
discount_p |
Propagation probability for |
ncvote_theta |
Weight of the plain vote in |
comm_r |
Scale |
ld_radius |
Radius for |
enrenew_depth |
Renewal radius for |
voterank_lambda |
Suppression factor for |
contraction_rho |
|
wks_alpha, wks_beta |
Degree and strength exponents for
|
renewed_threshold |
Diffusion-importance threshold for
|
kpath_k |
Maximum path length for |
kpath_len |
Maximum path length for |
epc_threshold |
Edge removal probability for |
epc_runs |
Number of percolation realizations for |
epc_seed |
Random seed for |
betweenness_delta |
Decay exponent for |
closeness_delta |
Distance exponent for |
gravity_mass |
Mass in |
gravity_radius |
Largest distance each gravity source reaches in
|
mdd_lambda |
Exhausted-degree weight for |
volume_radius |
Closed neighborhood radius for |
diffusion_q |
Multiplier between 0 and 1 for |
diffusion_steps |
Nonnegative integer horizon for
|
ds_beta |
Spreading rate for |
ds_mu |
Recovery rate for |
ds_steps |
Nonnegative integer horizon for |
cda_alpha |
Degree-versus-strength weight for |
icc_alpha |
Shortest-path multiplicity exponent for
|
exogenous_base |
Base for |
wlr_alpha |
Finite in-degree exponent for |
alr_h_mode |
H-index convention for |
grc_gamma |
Finite nonnegative regularization strength for
|
rwd_decay |
Finite first-arrival discount in |
rwd_node_weights |
Nonnegative starting weights for
|
linerank_aggregation |
LineRank endpoint aggregation: probability
(default) or weight. See |
bridging_steps |
Nonnegative bridging-capital walk horizon, default two. |
bridging_values |
Optional source-destination value matrix for
|
proximal_variant |
Proximal betweenness role: source (default),
target, sum, or union. See |
exf_alpha |
Modified Expected Force degree factor, default two, finite and greater than one. |
beta_direction |
BG-index orientation, positive (default) or negative.
See |
ninl_order |
Nonnegative NINL iteration count, default three. |
ninl_radius |
NINL hop radius, NULL for ceiling of mean path length.
See |
map_flow |
Map equation flow model, unrecorded (default) or recorded. |
map_convention |
Map equation coding convention, paper (default) or
infomap. See |
sr_prior |
SpectralRank diagonal prior, default zero; scalar or one
value per node. See |
mcgm_radius |
MCGM hop cutoff, default two; NULL includes all reachable nodes. |
mcgm_alpha |
MCGM coefficient, NULL for the published adaptive rule.
See |
dkgm_radius |
DKGM hop cutoff, default two as in the paper's printed
example; NULL or infinity includes all reachable nodes and "auto"
applies the paper's half-mean-distance rule with cograph rounding.
See |
nd_order |
Steps of neighbors summed by |
nd_decay |
Per-step decay for |
nd_mass |
Benchmark centrality summed by
|
ira_mass |
Node centrality allocated by |
ira_alpha |
Exponent on the |
ira_tol |
Stopping tolerance for |
ira_max_iter |
Iteration bound for |
iira_beta |
Spreading rate for |
iira_steps |
Iterations for |
hcc_delta |
Weight on a node's own degree in the extended degree
used by |
lhc_radius |
Radius of the ball |
tpr_alpha |
Jump probability of the trust-PageRank iteration used
by |
tpr_k |
Weight of the degree ratio in the trust value of
|
tpr_decay |
Attenuation factor of the similarity recursion used by
|
tpr_tol |
Convergence tolerance on the largest relative change of
either trust-PageRank recursion, a single positive number, default
|
tpr_max_iter |
Iteration bound for both trust-PageRank recursions, a
whole number of at least one, default 1000. Reaching it raises a
|
rsp_beta |
Inverse temperature of the randomized-shortest-paths
model used by |
rsp_cost |
How an edge weight becomes a traversal cost for
|
re_indexes |
Constituent indexes integrated by
|
re_negative |
Which of |
tna_network |
Logical or NULL. Umbrella switch that forces tna-style
conventions across all measures. |
psych_network |
Logical or NULL. Switch for signed psychometric
network conventions. |
shared_nodes |
Logical. Only used when |
... |
Additional arguments (currently unused) |
The following centrality measures are available:
Count of edges (supports mode: in/out/all)
Weighted degree (supports mode: in/out/all)
Shortest path centrality
Inverse distance centrality (supports mode: in/out/all)
Influence-based centrality
Random walk centrality (supports damping and personalization)
HITS authority score
HITS hub score
Maximum distance to other nodes (supports mode).
Distances use the raw edge weights whatever invert_weights
says, and hop counts with weighted = FALSE.
K-core membership (supports mode: in/out/all)
Burt's constraint (structural holes)
Local clustering coefficient (supports multiple types)
Harmonic centrality, the sum of inverse distances. It stays finite on disconnected graphs (supports mode: in/out/all)
Diffusion degree centrality. With
diffusion_method = "kandhway_kuri" it is the scaled degree of
the node plus the scaled degrees of its neighbors (supports mode:
in/out/all, lambda scaling)
Leverage centrality. Influence over neighbors based on relative degree differences (supports mode: in/out/all)
K-reach centrality. Number of nodes reachable within
k steps (supports mode: in/out/all)
Alpha centrality. Influence through paths, attenuated by length, with a unit exogenous contribution at every node (supports mode: in/out/all)
Bonacich power centrality. Influence based on connections to other influential nodes (supports mode: in/out/all)
Subgraph centrality. Participation in closed walks, with shorter walks weighted more heavily
Laplacian centrality with the local formula of Qi et
al. (2012). Matches NetworkX and centiserve::laplacian()
Load centrality. Share of shortest-path load passing through the node, with load split evenly at each branching point
Information centrality. Closeness based on electrical current flow (requires a connected graph)
Random-walk betweenness. Betweenness based on electrical current flow (requires a connected graph)
VoteRank. Influential spreaders selected by iterative
voting. The value is the election order rescaled so that the first
elected node scores 1 and the last scores 1/n
Percolation centrality. Shortest-path betweenness
weighted by the node states in states. With equal states it is
betweenness divided by (n-1)(n-2)
Radiality centrality (centiserve). Sum of (diam + 1 - d) normalized by n-1.
Lin's centrality. Reachable nodes squared divided by sum of distances.
Decay centrality. Sum of \delta^d over all nodes,
the node itself included, with \delta = decay_parameter.
Residual closeness. Sum of 1/2^d over
all nodes, the node itself included.
Dangalchev closeness. Same values as
residual_closeness.
Generalized closeness. Same formula as
decay, using decay_parameter.
Harary centrality. Sum of 1/d^2 for all reachable pairs.
Average distance (centiserve). Sum of distances / (n+1).
Barycenter centrality. 1 / sum of distances.
Wiener index. Total sum of shortest path distances from node.
Closeness vitality. Drop in Wiener index when node removed.
Total communicability. Row sums of matrix exponential.
Communicability betweenness. Fraction of communicability through each node.
Random walk centrality. Inverse sum of random walk distances (requires connected graph).
Stress centrality. Number of shortest paths through node.
Flow betweenness. Max-flow based betweenness.
Lobby index (h-index of neighborhood).
Graph entropy centrality. Entropy change on node removal.
Semi-local centrality. Triple-nested neighborhood sum.
ClusterRank. Clustering coefficient times neighbor degree sum.
Bottleneck centrality. Count of shortest path trees where node is critical.
Centroid value. Minimum f(v,i) across all nodes.
Maximum Neighborhood Component size.
Density of Maximum Neighborhood Component.
Topological coefficient. Shared neighbor ratio.
Bridging centrality. Betweenness times bridging coefficient.
Local bridging. (1/degree) times bridging coefficient.
Burt's effective size. Degree minus redundancy.
Diversity centrality. Shannon entropy of edge weight distribution.
Cross-clique connectivity. Count of cliques containing node.
Markov centrality. Inverse mean first passage time (requires connected graph).
Integration centrality. Distance-based influence.
Expected centrality. Sum of neighbor degrees.
Gil-Schmidt power index. Sum of 1/d normalized by n-1.
SALSA authority scores (directed graphs only).
LeaderRank. PageRank with ground node (directed graphs only).
Participation coefficient. Diversity of inter-community
connections (requires membership).
Within-module degree z-score. Intra-community
connectivity (requires membership).
Gateway coefficient. Inter-community brokerage weighted by
centrality (requires membership).
Normalized Shannon entropy of a node's hop-distance profile. It is 1 when the distances are spread evenly and 0 when all lie at one distance.
Growth exponent of the ball around a node
(slope of \ln B_i(r) on \ln r). Lower values mark more
influential nodes.
Entropy-weighted local dimension over boxes up to half the node's eccentricity. Higher values mark more influential nodes.
Mean degree of a node's neighbors (average neighbor degree); isolates score 0.
Drop in modularity when the node is removed
under a fixed partition. Positive values mark community hubs and
negative values mark bridges (requires membership).
Shapley value of the
node in the coverage games of Michalak et al. (2013): one-hop
coverage, shapley_k-neighbor coverage, and coverage within
shapley_cutoff hops. Values sum to the node count.
Mean bits needed to reach every other node along shortest paths without a map. Low values mark well-connected nodes.
Mean bits others need to find the node. High values mark hidden nodes.
Log rumor centrality on the node's BFS tree: log of the number of spreading orders that could start there.
Community size times intra-community
degree plus number of other communities touched times
inter-community degree (requires membership).
Drop in
the Shannon entropy of the degree (by mode) or betweenness
distribution when the node is deleted; signed, nats.
Shell index of the strength-based peeling with
asymmetric topological link weights, exponent s_shell_a.
Greedy seed-selection order
under degree discounting (discount_p) or unit discounting,
scored 1 for the first selected down to 1/n.
VoteRank with voters weighted by normalized
neighborhood coreness (ncvote_theta); election order scored
like voterank.
Links
weighted by the size of the community they reach; Gupta's scaled
intra/inter-degree combination (comm_r); base-2 entropy of the
link distribution over communities times degree share (all require
membership).
Silva-Costa estimator at ld_radius;
slope of the fuzzy ball, where higher values mark more influential
nodes; slope of the degree volume, where lower values mark more
important nodes.
Election orders of the
weighted, entropy-based (enrenew_depth) and degree-weighted
(voterank_lambda) VoteRank variants, scored like
voterank.
One minus the
agglomeration ratio after contracting the node with its neighbors;
the improved form adds the same score of its edges on the line graph
(contraction_rho).
Number of node pairs whose most likely two-way random-walk route passes through the node.
Farness minus mean neighbor farness. Lower values mark more central nodes.
Share of neighbor pairs linked through the node but not directly.
-\sum_{j \in N(i)} k_j \ln k_j. Lower values
mark more central nodes.
h-index over topological link weights
k_i k_j repeated k_j times.
Mean degree of the neighbors inside the ego network; degree minus effective size.
k-shell on (k^\alpha s^\beta)^{1/(\alpha
+ \beta)} after Garas' weight normalization (wks_alpha,
wks_beta).
k-core of the graph after removing links whose
diffusion importance is below renewed_threshold.
Number of shortest paths of length at most
kpath_k starting at the node.
Global efficiency of the subgraph induced on
the node's neighbors, the node itself removed. This differs from
igraph::local_efficiency(), which measures the distances
between those neighbors through the rest of the network.
Largest strength threshold whose s-core still contains the node; the k-core number when weights are absent.
Distance-weighted fragmentation of the network after deleting the node. Higher means a more disruptive removal.
Number of simple paths of length at most
kpath_len that the node lies on, endpoints included.
Edge percolated component: mean size of the node's
component over epc_runs bond-percolation realizations, as a
share of the network. A Monte Carlo estimate; epc_seed makes
it reproducible.
Betweenness with each separated pair
weighted by 1 / d(s,t).
Betweenness with the pair weight
(h(s,t) - 1)^{-\delta}
(betweenness_delta), where h(s,t) is the number of edges
on a shortest path.
Betweenness inside the node's own ego network.
\sum_j d_{ij}^{-\delta} / (n-1)
(closeness_delta).
Node truss number (k-2 triangles convention) and
mixed-degree shell threshold (mdd_lambda). Both use the
simple undirected skeleton; see centrality_truss.
Reciprocal-degree ratio, count of unconnected neighbor pairs, and count of triangle-supported relationships on the simple undirected skeleton.
Sum of degrees in the closed volume_radius-hop
neighborhood on the simple undirected skeleton.
Maximal clique centrality: sum of (|C|-1)! over
incident maximal cliques of size at least two. Costly; see
centrality_mcc for isolate and precision conventions.
Finite-horizon weighted outgoing walks:
\sum_{t=1}^{T}(qA)^t\mathbf{1}, with diffusion_q and
diffusion_steps. Distinct from diffusion degree.
Relative spectral-radius loss on vertex
deletion, evaluated by repeated eigendecomposition. Costly; see
centrality_dynamical_importance for zero-radius graphs.
Finite-time spreading score including
ds_beta, ds_mu and ds_steps; uses the simple
undirected skeleton.
Sum of focal-to-neighbor degree ratios on the simple undirected skeleton; the reciprocal of the bridging coefficient on nonisolated vertices.
One minus half the incident conductance
times effective-resistance sum. Weighted, componentwise and costly;
see centrality_resistance_curvature.
Sum of neighbors' neighborhood coreness; equivalently the squared simple adjacency times core numbers.
Gravity with the degree k-shell index as the mass at both
ends, default radius two; see centrality_dkgm.
Benchmark centrality plus its decayed sums
over non-backtracking walks of up to nd_order steps. See
centrality_neighbor_distance.
Steady state of a unit resource repeatedly reallocated to
neighbors in proportion to their ira_mass; conserved, so the
scores of a component sum to its size. Warns
cograph_no_converge where no steady state exists. See
centrality_ira.
The same recursion with each share scaled by
1-(1-\beta)^{k_i} for the iira_beta spreading rate,
run iira_steps times.
Decays geometrically, so only the order is meaningful. See
centrality_iira.
Local neighbor contribution: the cubed degree times the
binomial own-contribution factor (1-1/d_i)^{d_i-1} times the
neighbors' degree sum over n-1. Parameter-free; raw scores
depend on the whole graph's order. See centrality_lnc.
KED method: the degree times one plus the normalized
entropy of the neighbors' degrees times \exp(K_i/N) for the
neighbor-degree sum K_i and the whole graph's order
N. Parameter-free. See centrality_ked.
Hybrid characteristic centrality: the extended degree
\delta k_i+(1-\delta)\sum_{j\in N(i)}k_j over its maximum,
plus the E-shell peeling round in which the node leaves over the
number of rounds. Raw scores lie in [0,2] and are not
component-local. See centrality_hcc.
Extended hybrid characteristic centrality: the
closed-neighborhood sum of hcc, the focal node counted once.
See centrality_ehcc.
Lhc index: the degree-and-triangle-share influence
C(v)=\sum_{u\in\Phi(v)}k_u(1+TP(u))/d^2(uv) over the ball of
radius lhc_radius, summed over the open neighborhood. The
triangle share is normalized by TNTS=\sum_u NTS(u), three
times the number of distinct triangles, and is written as zero on a
triangle-free graph. Raw scores are not component-local. See
centrality_lhc.
Immediate effects centrality: the reciprocal mean length
of the influence sequences that end at a node,
(n-1)/\sum_{i\neq j}m_{ij} for the mean first passage times
M=(I-Z+EZ_{dg})\mathrm{diag}(1/c) of the influence chain
W=A/\mathrm{rowSums}(A) built with a_{ii}=1.
Direction-sensitive and costly (one eigenproblem and two dense
solves). NA at every node when the chain is reducible or the
graph has one node. Not the same measure as markov. See
centrality_iec.
Degree and importance of lines: the degree plus the share
of each incident line's importance I_e=(k_m-p-1)(k_n-p-1)/
(p/2+1) that the node's own degree claims,
k_i+\sum_{j\in\Gamma_i}I_{e_{ij}}(k_i-1)/(k_i+k_j-2), with
p the number of triangles on the line. Two-hop local and
component-local; never below the node's degree. See
centrality_dil.
Trust-PageRank: a damped PageRank whose split of
a node's score among its neighbors is the column-stochastic
trust-value T(i,j)=(1-k)s(i,j)/\sum_{l\in N_j}s(j,l)+
k\,d_i/\sum_{l\in N_j}d_l, with s the fixed point of SimRank
restricted to the lines of the graph. Scores sum to one when no node
is isolated. NA at every node of a component that has lines
but no triangle, where the similarity vanishes and the ratio is
undefined. Costly
(two fixed-point recursions over dense matrices). See
centrality_trust_pagerank.
Simple randomized shortest paths betweenness:
the expected number of visits a node receives over the Boltzmann
distribution on absorbing walks, summed over every ordered
source-target pair. rsp_beta interpolates between the
random-walk and shortest-path readings. Direction-sensitive,
component-local, and costly (one dense inverse). See
centrality_rsp_betweenness.
Normalized geometric mean of several index
distributions, the minimum-relative-entropy integration of
re_indexes; sums to one. See
centrality_relative_entropy.
Gravity with focal core-number and partner-degree masses, default radius three.
Sum of immediate neighbors' raw mixed gravitational centralities.
Sum of neighbors' raw k-shell gravity scores,
with gravity_radius applied around each neighbor.
Weighted degree and strength, adjusted by Barrat clustering,
plus weighted neighbor contributions; uses cda_alpha.
Closeness using distances divided by the
number of shortest paths raised to icc_alpha.
Contribution to all other nodes' base centrality,
measured by deletion. Selects a base using exogenous_base.
Exponential focal coreness times distance-discounted partner coreness (GSM).
Exponential degree-coreness influences with an adaptive distance exponent (H-GSM).
Exponential focal degree with partner degrees discounted by a global mean-degree distance exponent (IGSM).
Stationary scores with ground-node outgoing
weights determined by original in-degree and wlr_alpha.
PageRank on the line graph, aggregated at endpoints;
uses damping and linerank_aggregation.
Entropy of onward boundary degrees over all two-event transmission sequences.
Multi-characteristics gravity with degree, coreness and eigenvector masses; default radius two.
Outgoing Perron eigenvector with a unit-linked
ground node; sr_prior supplies optional diagonal information.
Smallest eigenvalue of each grounded symmetric
row-Laplacian; see centrality_controlrank.
Codelength saving on silencing a node, conditional on the supplied partition, flow model and coding convention.
Finite neighbor propagation of closed-neighborhood degree
volume; uses ninl_order and ninl_radius.
BG power shared by successors among predecessors;
beta_direction selects positive or negative orientation.
Betweenness in one-hop or two-hop ego networks times the original bridging coefficient.
Expected Force multiplied by
log degree with the scaling parameter exf_alpha.
First/last shortest-path intermediaries;
uses proximal_variant on the directed unweighted skeleton.
Counts four-edge nonbacktracking walks with each node at the middle, using original neighbor excess degrees.
Concentration of dyadic Burt constraints across contacts; isolates zero and single-contact nodes one.
Information-walk loss under single-entry deletion;
uses bridging_steps and bridging_values.
Weighted sum of discounted first arrivals
from random walks; uses rwd_decay and rwd_node_weights.
Reciprocal diagonal of the inverse
regularized weighted Laplacian, using grc_gamma.
Stationary scores with destination weights
determined by original H-indices using alr_h_mode.
A base data.frame with one row per node, in the input's node
order unless sort_by is given, and the columns:
node: character, the node labels (the index as a string
when the input carried no names)
One numeric column per requested measure, with a mode suffix for
the mode-aware measures (e.g., degree_in,
closeness_all); see list_centralities for which
measures carry a suffix. A measure that a tier supplied but that has
no value on this input is an all-NA column.
The community-aware
measures warn and return an all-NA column when
membership is missing, and the brokerage roles do the same on an
undirected graph. "relative_entropy" has no value when one of
its constituent indexes is zero at every node. Named in
measures or include, it then raises a
cograph_undefined_index error. Supplied by a tier, it gives a
cograph_undefined_measure warning and an all-NA column,
and the other measures of the tier are still computed.
"flow_betweenness" requires the igraph package. Without igraph,
naming it raises a cograph_needs_igraph error, and a tier gives
the same warning and all-NA column.
centrality(regulation_net)
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.