Date: November 29th, 2023
remstats
object are now correctly set after executing bind_remstats()
.remstats
failed to find the edgelist when reh
is ordinal.aomstats
failed when the computation method was pe
, the memory method was decay
and the first event at the start
index was not the first event at that time.Date: November 16th, 2023 Compatible with remify
version 3.2.0.
method
for managing simultaneous events. The default method
is "pt"
(per timepoint), where statistics are computed once for each unique timepoint in the edgelist. Alternatively, you can choose "pe"
(per event), where statistics are computed once for each unique event observed in the edgelist.same
statistic.subset
and method
object to the attributes of tomstats
and aomstats
objects.start
and stop
values changes slightly when method
is pt
. They now refer to the indices of the first unique timepoint and last unique timepoint for which the user wishes to calculate statistics. The meaning of start
and stop
remains unchanged when method
is pe
.consider_type
argument is now TRUE
for improved logical consistency.formula
attributes of aomstats
object.ccp
effect has been removed.Date: 25-10-23 Works with remify >= 3.1.0.
bind_remstats()
that occurs when combining two actor-oriented remstats
objects, one of which has receiver effects while the other does not.Date: 09-10-23 Works with remify >= 3.1.0.
tie()
statistic.tie()
statistic.dyad()
statistic as an alternative to the tie()
statistic.attr_dyads
) for the tie()
statistic, the previously used attr_data
object has been renamed to attr_actors
. This change ensures consistency and clarity in our terminology.Date: 06-10-2023 Works with remify >= 3.1.0.
Date: 05-09-2023 Works with remify >= 3.1.0.
bind_remstats()
now works for the actor-oriented modelDate: 28-08-2023 Works with remify >= 3.1.0.
remstats()
now works for ordinal event sequences.Date: 13-07-2023 Works with remify >= 3.1.0.
display_progress
argument for the tie-oriented model.Date: 06-07-2023 Works with remify >= 3.1.0.
unique
argument in triadic statisticsvar1.x.var2
to var1:var2
.spUnique
is deprecated. Instead sp(unique = TRUE)
can be used. This is done to be consistent with the new unique
argument in the directed triadic statistics.testthat
to tinytest
.Date: 01-07-2023
Works with remify >= 3.1.0.
bind_remstats()
remify::remify
riskset
argument)attributes
argument that exists in many of the function is renamed to attr_data
to avoid naming conflicts.scaling = "as.is"
option to scaling = "none"
.Date: 01-06-2023
IMPORTANT: remstats 3.1.0 works with remify >= 3.0.0. Make sure remify is updated. If earlier versions of remify are in use, remstats 3.1.0 will break down.
print()
, and summary()
methods.edgelist
argument is renamed to reh
.reh
must be an object of class remify
.id
column in attributes
object is renamed to name
.aomstats
output is reduced to the list of statistics (sender_stats and receiver_stats)tomstats
output is reduced to the array of statistics. The risk set is now an attribute. The adjmat is only an attribute if requested with “get_adjmat” (this is changed to save memory).Date: 16-Mar-23
tie_effects()
and actor_effects()
.display_progress
argument in aomstats()
(will be added to remstats()
and tomstats()
later.)Date: 08-Feb-23
Date: 05-Jan-23
Date: 22-Dec-21
IMPORTANT: remstats 3.0.0 works with remify >= 2.0.0. Make sure remify is updated. If earlier versions of remify are in use, remstats 3.0.0 will break down.
tie()
(see issue #29).Date: 21-Dec-21
Date: 16-Dec-21
degreeMin
and degreeMax
statistics for undirected events.ccp
statistic for undirected, dyadic events.verbose
argument that, when set to TRUE, outputs a progress update on the statistics computation.tomstats()
.event()
effect.tomstats
) for greater efficiency and less computation time.remstats()
function, which is a wrapper for aomstats()
and tomstats()
.aomstats()
function arguments that refer to the requested effects: sender_effects
(was rateEffects
) and receiver_effects
(was choiceEffects
).aomstats()
statistics
List output object to sender_stats
(was rate
) and receiver_stats
(was choice
).NEWS.md
file to track changes to the package.aomstats()
is added to compute statistics for the actor-oriented model. Effects for the rate-step and the choice-step of this model have to be specified separately, see the function’s documentation or the README.md.Brandes
is added to the memory
functionality. This refers to the exponential decay of the weight of past events, depending on the time that has past since the event occurred. The memoryValue
argument refers to the halftime parameter (see Brandes et al., 2009).remstats()
, which was the main function in the previous version, is changed to tomstats()
(because it computes statistics for the tie-oriented model, as opposed to the function aomstats()
, which computes statistics for the actor-oriented model).memory
and memoryValue
are added as arguments of tomstats()
and aomstats()
and removed from the separate statistic functions. This is because based on these memory settings, an adjacency matrix is computed. Based on this adjacency matrix, a lot of the endogenous effects are computed. While this increases the efficiency of the computation, memory effects can no longer be specified separately per endogenous variable.weight
column in the edgelist
that is supplied to tomstats()
or aomstats()
instead of to a separate argument in the separate statistic functions. This is because based on these weights, an adjacency matrix is computed that is used for the computation of the endogenous effects. (see previous point).effects
argument (tomstats()
) or rateEffects
and choiceEffects
argument (aomstats()
). Previously, this argument was called formula
.tomstats()
and aomstats()
are transformed back to their original ID’s, instead of to the ID’s used by remify::reh()
and internally, so that the user can more easily assess the computed statistics.lm()
. In the tie-oriented model and rate step of the actor-oriented model, an intercept term is assumed by default (unless the ordinal
argument is set to TRUE
). Alternatively, it can be explicitly specified by adding 1 to the terms of the effects formula or explicitly removed by adding -1 to the terms. The baseline()
specification is removed.FEtype()
(could previously be done with baseline(with_type = TRUE)
).with_type
in the endogenous statistic functions is renamed to consider_type
.attributes
object (previously named covariates
) can now be supplied to the main functions tomstats()
or aomstats()
directly, but can also still be specified in the separate functions for exogenous statistics.