as.data.frame.networkDynamic | R Documentation |
Creates a data.frame giving lists of edges and their activity spells corresponding to the networkDynamic
object. An observation window may be specified, and edge spells with values outside the window will be omitted or truncated/censored to the observation window bounds.
## S3 method for class 'networkDynamic'
as.data.frame(x, row.names = NULL, optional = FALSE, e =seq_along(x$mel),
start = NULL, end = NULL, active.default=TRUE,...)
x |
a networkDynamic object |
row.names |
ignored |
optional |
ignored |
e |
vector of edge ids that should be included in the output. |
start |
optional, numeric onset-censoring time: time at which the network was first observed. Edge spells with earlier onset times are given this as their onset time (and marked as left censored). See Details for the default values. |
end |
optional right-censoring time: time at which the network was last observed. Edge spells with larger terminus are given this as their terminus time (and marked as right censored). See Details for the default values. |
active.default |
logical indicating if edges with no defined activity spells should be considered active by default (and appear as a row in output) |
... |
possible additional arguments |
Exports the edge dynamics of a networkDynamic
object as a data.frame
giving a table of edges with their activity spells, including whether an activity spell was “censored” (onset or termination was not observed within the observation window).
If a net.obs.period
attribute exists, the maximal and minimal values will be used as censoring times by default. Censoring times can also be provided using the start
and end
arguments (which will override net.obs.period
), and censoring can be prevented by setting start=-Inf, end=Inf
. Edge spells that exceed the censoring bounds will have their onset and/or terminus time appropriately truncated to match and will be marked as censored. Spells which are fully outside the censoring bounds will be deleted and the corresponding rows will not be returned.
Note that the spell overlap rules mean that when edge spells are modeled as momentary events (onset==terminus), any events have onset==terminus==end of the the observation window will not be returned. This will likely be the case for networks created by the networkDynamic
converter function.
A data.frame
containing timed relational information
describing the networkDynamic object, having the following columns:
onset , terminus |
Onset and terminus times of an edge, respectively, if not censored, and left and right censoring times, if censored. |
tail , head |
Tail and head of the edge. |
onset.censored , terminus.censored |
TRUE if onset/terminus time is censored. |
duration |
the duration (terminus-onset) of each spell, evaluated after censoring |
edge.id |
the id of the edge the row corresponds to |
The output data.frame
will be sorted by edge.id, onset,terminus. This means that for most (non-multiplex) networks the group of spells for a specific edge will appear in sequential lines of output, but the overall output is not in a temporal order.
Vertex and attribute activity are not included in the returned data.frame. If an edge has multiple activity spells, it will have multiple rows in the output. When output in censored, it may not match values returned by get.edge.activity
See Also as get.edge.activity
,get.vertex.activity
net <-network.initialize(3)
net[1,2]<-1;
net[2,3]<-1;
## censoring
activate.edges(net,onset=1,terminus=Inf,e=1)
activate.edges(net,onset=2,terminus=3,e=2)
activate.vertices(net, onset=1, terminus=Inf, v=1)
as.data.frame(net)
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.