NEWS

---

PLMIX 2.1.2

PLMIX 2.1.1 (2019-09-04)

NEW FEATURES

• Added a new dataset called 'd_gaming'.

DOCUMENTATION

• Removed some typos from the manual.

PLMIX 2.1.0 (2019-08-03)

MAJOR CHANGES

• The pi_inv argument (formerly a numeric N*K matrix with the observed top-ordering) has been redefined as an object of class 'top_ordering' or an object that can be coerced with the new function as.top_ordering.
• gibbsPLMIX(...):
• [::] added the element 'objective' in the output list representing the objective function values (that is the kernel of the log-posterior distribution) sampled in the saved MCMC iterations.
• mapPLMIX(...):
• [::] recoded default value of the argument 'plot_objective' from TRUE to FALSE.
• make_complete(...), make_partial(...), paired_comparisons(...), rank_ord_switch(...) and rank_summaries(...):
• [::] renamed argument 'format' as 'format_input'.
• [::] deleted default value ("ordering") for the former argument 'format'.
• rPLMIX(...):
• [::] renamed argument 'format' as 'format_output'.
• selectPLMIX(...):
• [::] renamed argument 'post_summary' as 'post_est'.
• 'd_apa', 'd_carconf', 'd_dublinwest', 'd_german', 'd_nascar', 'd_occup' and 'd_rice'
• [::] appended S3 class 'top_ordering' to the supplied top-ordering datasets.
• [::] renamed columns of the supplied top-ordering datasets.

MINOR CHANGES

• fixed a CRAN error due to the cleanEX(...) procedure in the R CMD check.
• mapPLMIX(...):
• [::] reversed order between the 'P_map' and 'W_map' elements in the output list.
• rank_ord_switch(...):
• [::] added names to the columns of the output matrix.
• unit_to_freq(...):
• [::] added name 'freq' to the last column of the output matrix.

NEW FEATURES

• Added the new membership function is.top_ordering(...) that verifies whether the supplied data satisfy the conditions to be a top-ordering dataset (various object classes are accepted for the input data).
• Added the new coercion function as.top_ordering(...) that attempts to coerce the supplied data into an object of S3 class 'top_ordering' (various object classes and forms are accepted for the input data).
• Appended S3 class 'mpPLMIX' to the list returned by mapPLMIX(...) and mapPLMIX_multistart(...).
• Added generic S3 class methods for objects of class 'mpPLMIX' named as print.mpPLMIX(...), summary.mpPLMIX(...) and plot.mpPLMIX(...).
• Appended S3 class 'gsPLMIX' to the list returned by gibbsPLMIX(...).
• Added generic S3 class methods for objects of class 'gsPLMIX' named as print.gsPLMIX(...), summary.gsPLMIX(...) and plot.gsPLMIX(...).
• Added the new function gsPLMIX_to_mcmc(...) to coerce an object of class 'gsPLMIX' into an object of class 'mcmc' as defined by the function as.mcmc(...) of the 'coda' package.
• Added two further datasets called 'd_occup' and 'd_rice'.
• Added packages 'coda', 'ggmcmc', 'ggplot2', 'radarchart', 'prefmod', 'rankdist', 'StatRank', 'pmr', 'PlackettLuce', 'utils' and 'reshape2' in Imports.

DOCUMENTATION

• The manual has been globally reviewed and improved accordingly with the new changes and the additional implementations described in NEW FEATURES.
• label_switchPLMIX(...):
• [::] added the citation Papastamoulis, P. (2016).

PLMIX 2.0.1 (2018-05-02)

MINOR CHANGES

• fixed former warnings using if() with conditions of length greater than one. We have also checked the current version with _R_CHECK_LENGTH_1_CONDITION_ set to true.
• gibbsPLMIX(...):
• [::] colnames added to output values labelled as 'P' and 'W'.
• mapPLMIX(...):
• [::] dimnames change for output value labelled as 'P_map'.
• [::] names added to output value labelled as 'W_map'.
• rank_summaries(...):
• [::] format and label change for the former output value labelled as 'missing_pos' now labelled as 'na_or_not'.
• [::] output values labelled as 'mean_rank' and 'marginals' now contain dimnames.

PLMIX 2.0 (2018-02-05)

BUG FIXES

• mapPLMIX(...)
• random_start(...)
• rank_summaries(...)

MAJOR CHANGES

• bicPLMIX(...):
• [::] replaced former argument 'weights' as 'G'.
• [::] added default values for the arguments 'ref_known' and 'ref_vary'.
• [::] added the reference Schwarz (1978).
• gibbsPLMIX(...):
• [::] added the 'centered_start' argument for the random generation of the starting values, similarly to the mapPLMIX(...) and mapPLMIX_multistart(...) functions.
• label_switchPLMIX(...):
• [::] added new function to perform label switching adjustment of the posterior MCMC samples, formerly implemented by both the ppcheckPLMIX(...) and the ppcheckPLMIX_cond(...) functions.
• mapPLMIX(...):
• [::] modified the default value of the argument 'eps' in accordance with that of the related mapPLMIX_multistart(...) function.
• [::] modified the default value of the argument 'plot_objective'.
• [::] dropped object 'Rho_map' from the output list.
• [::] renamed object 'z' of the output list as 'z_hat'.
• [::] renamed object 'classification' of the output list as 'class_map'.
• [::] 'bic' value is computed only for the default flat priors case, otherwise NULL is returned.
• mapPLMIX_multistart(...):
• [::] recoded default value of the argument 'init'.
• ppcheckPLMIX(...) and ppcheckPLMIX_cond(...):
• [::] removed former arguments 'MAPestP', 'MAPestW', 'ls_adj' and 'adj_post_sample' needed for the label switching adjustment, which is currently implemented by the novel label_switchPLMIX(...) function.
• [::] removed 'final_sampleP' and 'final_sampleW' objects from the output.
• rank_summaries(...):
• [::] renamed the argument and output object 'pairedcomparisons' as 'pc'.
• [::] renamed the output object 'missing_positions' as 'missing_pos',
• rPLMIX(...):
• [::] renamed former 'rankingoutput' argument with 'format'.
• selectPLMIX:
• [::] removed argument 'log_lik'.

MINOR CHANGES

• ppcheckPLMIX(...) and ppcheckPLMIX_cond(...):
• [::] added some messages.
• "G=" replaced with "G_".
• make_partial(...):
• [::] renamed argument 'probcensoring' as 'probcens'.

DOCUMENTATION

• The manual has been globally reviewed and improved with additional implementation details and some new examples.
• Added new examples for the three data objects 'd_apa', 'd_carconf' and 'd_dublinwest' showing how to extract the subset of complete orderings from the whole dataset.
• Updated reference Mollica and Tardella (2016) as Mollica and Tardella (2017).
• d_german dataset: modified the citation Croon (1989).
• mapPLMIX(...) and mapPLMIX_multistart(...):
• [::] added the citation Gormley and Murphy (2006).
• ppcheckPLMIX(...) and ppcheckPLMIX_cond(...):
• [::] removed example of label switching adjustment.
• selectPLMIX(...):
• [::] added Details section.
• [::] renamed output objects 'fitting_measures', 'effective_number_of_parameters' and 'selection_criteria' as 'fitting', 'penalties' and 'select_criteria'.

CITATION INFO

• Added bibTex entries in the file CITATION.
• Updated reference Mollica and Tardella (2016) as Mollica and Tardella (2017) in the CITATION info.

NEW FEATURES

• Removed package 'combinat' from Imports.
• Changed order in the namespace list (former: foreach abind rcdd FSA).