bccm(): fixed bug that returned wrongly filled omegaBlock matrix. The bug should not have affected the model fit itself.new features:
extended testing scripts of package
bug fixes:
logl() and rghype(): fixed bug that treated as a hypergeometric model a ghype where all odds are equal to 0. Throws error instead.
compute_xi(): Fixed a bug that introduced spurious stubs for nodes with zero degrees when creating a Xi matrix without selfloops
link_significance(): fixed bug when under=TRUE, updated implementation to improve performance
ghype(): fixed computation of degrees of freedom for full model where xi has zero values
link_significance(): When calling with under = FALSE and pval = FALSE was returning phi(0) = 1 instead of Pr(x>0)function renaming:
new features
updated documentation for endogenous statistics functions
introduced a new function get_zero_dummy() that generates the dummy variable to be used together with endogenous statistics functions.
added testing with test_that
check_specs(): recognize bipartite graphs
addressed issue #15: bccm now names coefficients according to label names. E.g.: coefficient for the propensities between block ‘red’ and ‘blue’ are now named ‘red<->blue’.
bug fixes:
scm(): fixed df for bipartite graphs
linkSignificance(): fixed issue for bipartite graphs
bccm(): fixed issue when fitting to bipartite graphs
bccm(): fixed issue arising when passing a list of identical labels (all vertices in one block)
NEWS.md file to track changes to the package.check_specs() that threw an error when the function was called without additional parameters.bccm() to improve its speed when passing a large number of labels.nrmSelection(), createPredictors, and nrmChoose() to nrm_selection(), create_predictors, and nrm_choose() respectively, to standardise functions’ names.nrm_selection() and nrm_choose() to use ‘pbmclapply’ and show a progress bar.