Skip to contents

Topological overlap: the share of a vertex's ties that survive from one time bin to the next, normalised by the geometric mean of its degrees in the two bins. Averaged over a vertex's transitions it is that vertex's persistence; averaged again over vertices it is the network's temporal correlation coefficient (Tang et al., 2010).

Usage

persistence(
  dn,
  scope = c("pertime", "node", "overall"),
  sessions = c("bounded", "collapse", "separate"),
  start = NULL,
  end = NULL,
  step = NULL,
  window = NULL,
  plot = FALSE
)

Arguments

dn

A temporal network from dynet().

scope

"pertime", the default, for one row per vertex per transition; "node" for each vertex's mean over its transitions; "overall" for the temporal correlation coefficient, one number per session.

sessions

How to treat sessions, as in centrality_series().

start, end, step, window

The measurement grid, as in snapshots(). A transition is a consecutive pair of bins on that grid, so the grid decides what "persisted" means; widening step makes persistence easier.

plot

Whether to draw the result as well as return it. Drawing is a side effect in the manner of graphics::hist(): the verb still returns its tidy table, invisibly when it has drawn.

Value

A dynet_metric. Under scope = "pertime" it is level = "node" with columns time, node, measure and value, one row per vertex per transition, where time is the earlier bin of the pair; the final bin opens no transition and so contributes no rows at all, rather than a row whose value is structurally undefined. Under scope = "node" it is level = "node" with node, measure and value, one row per vertex. Under scope = "overall" it is level = "graph" with measure and value, one row per session. A leading session column is present under sessions = "separate". measure is "topological_overlap", "average_topological_overlap" and "temporal_correlation" respectively, and value lies in [0, 1] at every scope. Print it, summary() it, plot() it, or take the plain frame with as.data.frame().

Details

The measure is defined on undirected neighbourhoods, so a directed network is read as a contact network here: each layer is folded onto its transpose before the overlap is taken. Weights and spell counts are ignored – a pair is tied in a bin or it is not – and loops are excluded.

A vertex isolated in either bin of a transition scores zero rather than a missing value. That is a substantive claim, not an arithmetic accident: the ratio is genuinely 0/0, and zero is the convention teneto uses. A reader who wants isolation and genuine turnover distinguished should read the degree alongside, which centrality_series() gives.

Conditions

Errors: dynet_empty_result when the grid yields fewer than two bins, since a single bin opens no transition and the measure is undefined; dynet_no_sessions when sessions = "separate" is asked of a network with no session column; and dynet_bad_input when dn is not a dynet.

References

Tang, J., Scellato, S., Musolesi, M., Mascolo, C., & Latora, V. (2010). Small-world behavior in time-varying graphs. Physical Review E, 81(5), 055101(R). doi:10.1103/PhysRevE.81.055101

Nicosia, V., Tang, J., Mascolo, C., Musolesi, M., Russo, G., & Latora, V. (2013). Graph metrics for temporal networks. In P. Holme & J. Saramaki (Eds.), Temporal Networks (pp. 15-40). Springer. doi:10.1007/978-3-642-36461-7_2

Clauset, A., & Eagle, N. (2007). Persistence and periodicity in a dynamic proximity network. DIMACS Workshop on Computational Methods for Dynamic Interaction Networks.

See also

similarity(), which compares whole edge sets between every pair of bins and answers a different question, and turnover() for the complementary view of what changed.

Examples

dn <- dynet(school_contacts)
kept <- persistence(dn)
kept
#> # Neighbourhood persistence per transition (node-level)
#> # 14 vertices | 21 time points, 1 per bin | time in step
#> # 1 keeps every neighbour, 0 keeps none; an isolated vertex scores 0
#>  time  node             measure     value
#>     0   Ana topological_overlap 0.0000000
#>     0   Ben topological_overlap 0.0000000
#>     0  Cara topological_overlap 1.0000000
#>     0   Dan topological_overlap 1.0000000
#>     0   Eve topological_overlap 0.4082483
#>     0  Finn topological_overlap 0.0000000
#>     0  Gita topological_overlap 0.0000000
#>     0  Hugo topological_overlap 0.0000000
#>     0  Iris topological_overlap 0.4082483
#>     0 Jonas topological_overlap 1.0000000
#>     0  Kira topological_overlap 0.7071068
#>     0   Leo topological_overlap 0.0000000
#> # 282 more rows. summary() aggregates them; plot() draws them.

per_node <- persistence(dn, scope = "node")
per_node
#> # Neighbourhood persistence per vertex (node-level)
#> # 14 vertices | time in step
#> # 1 keeps every neighbour, 0 keeps none; an isolated vertex scores 0
#>   node                     measure     value
#>    Ana average_topological_overlap 0.3736034
#>    Ben average_topological_overlap 0.3999443
#>   Cara average_topological_overlap 0.4203873
#>    Dan average_topological_overlap 0.4037204
#>    Eve average_topological_overlap 0.3901184
#>   Finn average_topological_overlap 0.2970579
#>   Gita average_topological_overlap 0.2680499
#>   Hugo average_topological_overlap 0.4283606
#>   Iris average_topological_overlap 0.3508067
#>  Jonas average_topological_overlap 0.4012369
#>   Kira average_topological_overlap 0.5274689
#>    Leo average_topological_overlap 0.3489197
#> # 2 more rows. summary() aggregates them; plot() draws them.

overall <- persistence(dn, scope = "overall")
overall
#> # Temporal correlation coefficient (graph-level)
#> # time in step
#> # 1 keeps every neighbour, 0 keeps none; an isolated vertex scores 0
#>               measure     value
#>  temporal_correlation 0.3894722