NEWS

---

MSmix 1.0.2 (2024-06-15)

BUG FIXES

• subsetting implementation with 'subset' argument in the case of numerical row indeces.
• bootstrapMSmix(...): [::] computation of the itemwise CIs for the components of the estimated consensus rankings in the case of tied relative frequencies.

MINOR CHANGES

• bootstrapMSmix(...): [::] added new component 'conf_level' in the output list. [::] added error message for the presence of rows in the 'rankings' argument having all NA entries.
• data_augmentation(...): [::] added error message for the presence of rows in the 'rankings' argument having all NA entries. [::] removed argument 'subset'.
• data_censoring(...): [::] argument 'probcens' renamed as 'probs'.
• data_completion(...): [::] argument 'reference_rho' renamed as 'ref_rho'. [::] added error message for the presence of rows in the 'rankings' argument having all NA entries. [::] added error message for the presence of some NA entries in the 'ref_rho' argument.
• data_conversion(...): [::] added error message for the presence of rows in the 'rankings' argument having all NA entries.
• data_description(...): [::] added error message for the presence of rows in the 'rankings' argument having all NA entries.
• fitMSmix(...): [::] added error message for the presence of rows in the 'rankings' argument having all NA entries.
• likMSmix(...): [::] added error message for the presence of rows in the 'rankings' argument having all NA entries.
• plot.data_descr(...): [::] added new argument 'marg_by' to report marginal distribution in the heatmap either by item or by rank. [::] changed scale and colors in the legend of the heatmap. [::] unified the colors of the items in the pictogram, ecdf and bubble plots.
• spear_dist(...): [::] added error message for the presence of some NA entries in the 'rankings' or 'rho' arguments.
• summary.emMSmix(...): [::] changed element 'MAP_distr' in the output list from relative to percentage distribution in the case of multiple mixture components, otherwise set as NULL.
• character variables converted into factors in the 'ranks_sports' dataset.

NEW FUNCTIONS

• added the print method print.bootMSmix(...) for the S3 class 'bootMSmix'.

NEW FEATURES

• added new argument 'subset' in bootstrapMSmix(...) and data_conversion(...).
• fitMSmix(...): [::] argument 'tune_theta' renamed as 'theta_tune'. [::] added new argument 'theta_tol'. [::] added new element 'best_log_lik' in the 'mod' output list. [::] element 'max_log_lik_part' renamed as 'best_log_lik_part' in the 'mod' output list.
• added new dataset 'ranks_horror'.
• added covariates in the 'ranks_antifragility' dataset.
• added packages 'magrittr' and 'scales' in Imports.
• removed package 'stringr' from Imports.
• moved package 'dplyr' from Imports to Depends.

DOCUMENTATION

• added package vignette
• the manual has been globally reviewed and updated with the new implementation details.

MSmix 1.0.1 (2024-04-09)

BUG FIXES

• fitMSmix(...) [::] stopping rule of the EM algorithm adjusted to avoid ERROR in CRAN checks for MacOS-arm64.

MAJOR CHANGES

• bootstrapMSmix(...): [::] replaced former arguments 'n_clust', 'z_hat', 'rho_mle' and 'theta_mle' with 'object' to exploit the S3 class "emMSmix". [::] added new argument 'parallel' for parallel execution of model fit in the case of partial rankings.
• confintMSmix(...): [::] replaced former arguments 'n_items', 'theta_mle', 'weights_mle' and 'z_hat' with 'object' to exploit the S3 class "emMSmix". [::] indication of the function validity for full rankings only.
• spear_dist_distr(...): [::] new implementation to allow computation also for n_items>170 and return the approximate distribution over a fixed grid of Spearman distance values to limit computational burden.

NEW FUNCTIONS

• data_censoring(...) for converting full rankings into partial rankings.

NEW FEATURES

• added package 'gmp' in Imports.

DOCUMENTATION

• the manual has been globally reviewed and updated with the new implementation details.