igraph's handling of
NAs #1821.new functions submodel_adjacencies_to_igraph() and plot_submodel_network()
for representing submodel relations as graphs, and for plotting them as networks.
argument scoretype is removed from frscore() and frscored_cna().
rean_cna() and frscored_cna() gain a new argument quiet.
minor internal changes for compatibility with cna v.4.0.0
cna package.rean_cna() and frscore() display progress tickers.
frscore() and frscored_cna() are no longer sensitive to locale (due to
collation differences) when
number of unique model types exceeds maxsols.
unless normalize = "none", norm.score column filled with zeroes is added to the output of frscore()
and frscored_cna() when maximum score obtained by a model is zero.
frscore() and frscored_cna() return an adjacency matrix where
adjacencies represent submodel relations between the scored model types.
frscored_cna() and frscore() return how they were called.
frscored_cna() and rean_cna() accept output = "msc", in which case only mscs are calculated in, and collected
from the reanalysis series.
A new function causal_submodel() is introduced. causal_submodel() is like cna::is.submodel(), but checks that all
causal relevance ascriptions, rather than ascriptions of direct causation only,
of one model are contained in another model.
frscore() is updated to use causal_submodel() instead of cna::is.submodel() to compare models when calculating fr-scores.
frscored_cna() and frscore() gain an argument comp.method that
allows the user to decide whether fr-scores are calculated using causal_submodel() (default) or cna::is.submodel().
frscore() gains a new argument dat, to be used to determine
admissible factor values when processing multi-valued models using
comp.method = "causal_submodel".
frscored_cna() and rean_cna() gain a new argument n.init that
controls the maximum number of csfs that are calculated in each analysis.
This replaces ncsf in rean_cna(), ncsf is deprecated.
The scoretype argument in frscore() and frscored_cna() is deprecated
ahead of removal in the next version.
The unadjusted (non-normalized) fr-score of each model type is now always
included as the score column in the results returned by
frscore() and frscored_cna(). A normalized score
is included as a column named norm.score if normalize = "truemax" or
normalize = "idealmax".
A list containing a breakdown of each model's unadjusted fr-score is now
always included in the output of frscore() and frscored_cna().
The verbose argument now only controls
whether it is printed (verbose = TRUE) or not (verbose = FALSE).
Print methods for frscore()and frscored_cna() have been updated
to reflect the change in verbose behavior, and so that objects
created by previous versions of frscore() and frscored_cna() are
still printed correctly.
frscored_cna() documentation that started throwing an error due to changes in the cna package, and stops frscore() from displaying a confusing warning when the number of solution types exceeds the value of the argument maxsols.