- removed dependency from
extraDistr(orphaned package):
- introduced helper functions for handling multivariate hypergeometric density and sampling:
dmvhyper_baseandrmvhyper_base.
-
fixed reversed dependency issues for dplyr:::id "no visible binding for global variable 'id'" errors.
-
bug fixes:
bccm(): fixed bug that returned wrongly filled omegaBlock matrix. The bug did not affect the model fit itself.
- new features:
- extended testing scripts of package
- bug fixes:
logl()andrghype(): 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 selfloopslink_significance(): fixed bug when under=TRUE, updated implementation to improve performanceghype(): fixed computation of degrees of freedom for full model where xi has zero values
- bug fixes:
link_significance(): When calling with under = FALSE and pval = FALSE was returning phi(0) = 1 instead of Pr(x>0)
- function renaming:
linkSignificance()-->link_significance()ComputeXi()-->compute_xi()
- 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 graphslinkSignificance(): fixed issue for bipartite graphsbccm(): fixed issue when fitting to bipartite graphsbccm(): fixed issue arising when passing a list of identical labels (all vertices in one block)
- Added a
NEWS.mdfile to track changes to the package. - Fixed a bug in
check_specs()that threw an error when the function was called without additional parameters. - Modified
bccm()to improve its speed when passing a large number of labels. - Renamed
nrmSelection(),createPredictors, andnrmChoose()tonrm_selection(),create_predictors, andnrm_choose()respectively, to standardise functions' names. - Updated
nrm_selection()andnrm_choose()to use 'pbmclapply' and show a progress bar. - Added compatibility to 'texreg' to visualise the output of nrm as a table of parameters.