.github/workflows/test-coverage.yaml

@@ -1,4 +1,4 @@
-# Workflow derived from https://github.com/r-lib/actions/tree/master/examples
+# Workflow derived from https://github.com/r-lib/actions/tree/v2/examples
 # Need help debugging build failures? Start at https://github.com/r-lib/actions#where-to-find-help
 on:
   push:
@@ -8,37 +8,55 @@ on:
 
 name: test-coverage
 
+permissions: read-all
+
 jobs:
   test-coverage:
     runs-on: ubuntu-latest
     env:
       GITHUB_PAT: ${{ secrets.GITHUB_TOKEN }}
+
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
+
       - uses: r-lib/actions/setup-r@v2
         with:
           use-public-rspm: true
+
       - uses: r-lib/actions/setup-r-dependencies@v2
         with:
-          extra-packages: any::covr
+          extra-packages: any::covr, any::xml2
           needs: coverage
+
       - name: Test coverage
         run: |
-          covr::codecov(
+          cov <- covr::package_coverage(
             quiet = FALSE,
             clean = FALSE,
-            install_path = file.path(Sys.getenv("RUNNER_TEMP"), "package")
+            install_path = file.path(normalizePath(Sys.getenv("RUNNER_TEMP"), winslash = "/"), "package")
           )
+          covr::to_cobertura(cov)
         shell: Rscript {0}
+
+      - uses: codecov/codecov-action@v4
+        with:
+          # Fail if error if not on PR, or if on PR and token is given
+          fail_ci_if_error: ${{ github.event_name != 'pull_request' || secrets.CODECOV_TOKEN }}
+          file: ./cobertura.xml
+          plugin: noop
+          disable_search: true
+          token: ${{ secrets.CODECOV_TOKEN }}
+
       - name: Show testthat output
         if: always()
         run: |
           ## --------------------------------------------------------------------
-          find ${{ runner.temp }}/package -name 'testthat.Rout*' -exec cat '{}' \; || true
+          find '${{ runner.temp }}/package' -name 'testthat.Rout*' -exec cat '{}' \; || true
         shell: bash
-      - name: Upload test results
+
+      - name: Upload failures, if any
         if: failure()
-        uses: actions/upload-artifact@v3
+        uses: actions/upload-artifact@v4
         with:
           name: coverage-test-failures
           path: ${{ runner.temp }}/package

BayesMallows.Rproj

@@ -1,4 +1,5 @@
 Version: 1.0
+ProjectId: 6e0ef04a-160c-463b-bc1b-40b5326f6cb9
 
 RestoreWorkspace: No
 SaveWorkspace: No

DESCRIPTION

@@ -1,7 +1,7 @@
 Package: BayesMallows
 Type: Package
 Title: Bayesian Preference Learning with the Mallows Rank Model
-Version: 2.1.0
+Version: 2.2.3.9000
 Authors@R: c(person("Oystein", "Sorensen",
                     email = "oystein.sorensen.1985@gmail.com",
                     role = c("aut", "cre"),
@@ -46,12 +46,11 @@ BugReports: https://github.com/ocbe-uio/BayesMallows/issues
 License: GPL-3
 Encoding: UTF-8
 LazyData: true
-RoxygenNote: 7.3.1
+RoxygenNote: 7.3.2
 Depends: R (>= 3.5.0)
 Imports: Rcpp (>= 1.0.0),
   ggplot2 (>= 3.1.0),
   Rdpack (>= 1.0),
-  igraph (>= 1.2.5),
   sets (>= 1.0-18),
   relations (>= 0.6-8),
   rlang (>= 0.3.1)

NAMESPACE

@@ -28,6 +28,7 @@ export(assess_convergence)
 export(assign_cluster)
 export(burnin)
 export(compute_consensus)
+export(compute_exact_partition_function)
 export(compute_expected_distance)
 export(compute_mallows)
 export(compute_mallows_mixtures)
@@ -52,6 +53,7 @@ export(set_compute_options)
 export(set_initial_values)
 export(set_model_options)
 export(set_priors)
+export(set_progress_report)
 export(set_smc_options)
 export(setup_rank_data)
 export(update_mallows)

NEWS.md

@@ -1,3 +1,62 @@
+# BayesMallows (development versions)
+
+* Making plotting functionality compatible with future versions of ggplot2.
+
+# BayesMallows 2.2.3
+
+* Fixed bug, clus_thinning argument did not work in the case of a single 
+  cluster. Thanks to Timothy Lee for pointing this out. Issue #423 on GitHub.
+
+# BayesMallows 2.2.2
+
+* An error in compute_mallows_loglik when the number of clusters is more than
+  one has been corrected. Thanks to Marta Crispino.
+
+# BayesMallows 2.2.1
+
+* Skipping a unit test which failed on CRAN's M1 Mac builder.
+
+# BayesMallows 2.2.0
+
+* For initialization of latent ranks when using pairwise preference data, all
+  topological sorts are now generated in random order.
+* The SMC function now check for consistency with previous latent ranks for 
+  existing users also when data arrive in the form of pairwise preferences.
+* A function compute_exact_partition_function() is now added, which returns the
+  logarithm of the exact partition function for Cayley, Hamming, and Kendall
+  distance.
+* Fixed a bug in the Ulam distance. Thanks for Marta Crispino for discovering 
+  it.
+* Fixed a bug in SMC algorithm for pairwise preference data, where the proposal
+  distribution incorrectly was assumed to be uniform.
+* It is now possible to report progress of MCMC more flexibly using 
+  compute_mallows() or compute_mallows_mixtures(). The old argument "verbose"
+  which by default reported every 1000'th iteration has been replaced by an
+  argument "progress_report" which can be set by calling set_progress_report().
+  The latter allows setting the interval between reports. This is particularly
+  useful for big data, where running 1000 iterations may take very long time.
+* Fixed a bug which caused inconsistent partial rank data to be retained from
+  previous timepoints when existing users update their preferences.
+* Arguments random and random_limit to setup_rank_data() have been removed. A 
+  new argument max_topological_sorts has been added instead, which captures all 
+  previous use cases, but also allows the user to specify the number of 
+  topological sorts to generate. This makes it useful also with a relatively 
+  large number of items, while it previously would be computationally unfeasible
+  for anything more than 8-9 items.
+* Argument shuffle_unranked to setup_rank_data() has been removed. If there are 
+  unranked items they will now always be shuffled. For reproducibility, set the
+  random number seed.
+* SMC Mallows with pairwise preference data now allows different initially
+  values for the augmented rankings across the particles. This is obtained by
+  generating (a subset of) all topological sorts consistent with the transitive
+  closure for the user, and sampling from these. Can be set with the 
+  max_topological_sorts argument to set_smc_options().
+
+# BayesMallows 2.1.1
+
+* Fixed gcc-UBSAN issue happening when compute_mallows_sequentially() is run
+  without user IDs specified.
+
 # BayesMallows 2.1.0
 
 * The SMC method update_mallows() now supports pairwise preferences, both new 

R/RcppExports.R

@@ -5,6 +5,10 @@ abind <- function(x, y) {
     .Call(`_BayesMallows_abind`, x, y)
 }
 
+all_topological_sorts <- function(prefs, n_items, maxit, save) {
+    .Call(`_BayesMallows_all_topological_sorts`, prefs, n_items, maxit, save)
+}
+
 #' Asymptotic Approximation of Partition Function
 #'
 #' Compute the asymptotic approximation of the logarithm of the partition function,
@@ -68,8 +72,8 @@ rmallows <- function(rho0, alpha0, n_samples, burnin, thinning, leap_size = 1L,
     .Call(`_BayesMallows_rmallows`, rho0, alpha0, n_samples, burnin, thinning, leap_size, metric)
 }
 
-run_mcmc <- function(data, model_options, compute_options, priors, initial_values, pfun_values, pfun_estimate, verbose = FALSE) {
-    .Call(`_BayesMallows_run_mcmc`, data, model_options, compute_options, priors, initial_values, pfun_values, pfun_estimate, verbose)
+run_mcmc <- function(data, model_options, compute_options, priors, initial_values, pfun_values, pfun_estimate, progress_report) {
+    .Call(`_BayesMallows_run_mcmc`, data, model_options, compute_options, priors, initial_values, pfun_values, pfun_estimate, progress_report)
 }
 
 run_smc <- function(data, new_data, model_options, smc_options, compute_options, priors, initial_values, pfun_values, pfun_estimate) {

R/compute_exact_partition_function.R

@@ -0,0 +1,26 @@
+#' @title Compute exact partition function
+#'
+#' @description For Cayley, Hamming, and Kendall distances, computationally
+#' tractable functions are available for the exact partition function.
+#'
+#' @param alpha Dispersion parameter.
+#' @param n_items Number of items.
+#' @param metric Distance function, one of "cayley", "hamming", or "kendall".
+#'
+#' @return The logarithm of the partition function.
+#' @export
+#'
+#' @references \insertAllCited{}
+#'
+#' @example inst/examples/compute_exact_partition_function_example.R
+#' @family partition function
+compute_exact_partition_function <- function(
+    alpha, n_items,
+    metric = c("cayley", "hamming", "kendall")) {
+  metric <- match.arg(metric, c("cayley", "hamming", "kendall"))
+  validate_integer(n_items)
+  validate_positive(n_items)
+  validate_positive(alpha)
+
+  get_partition_function(alpha, n_items, metric, NULL)
+}

R/compute_mallows.R

@@ -14,8 +14,8 @@
 #'   is the latent consensus ranking, \eqn{Z_{n}(\alpha)} is the partition
 #'   function (normalizing constant), and \eqn{d(r,\rho)} is a distance function
 #'   measuring the distance between \eqn{r} and \eqn{\rho}. We refer to
-#'   \insertCite{vitelli2018;textual}{BayesMallows} for further details of the Bayesian
-#'   Mallows model.
+#'   \insertCite{vitelli2018;textual}{BayesMallows} for further details of the
+#'   Bayesian Mallows model.
 #'
 #'   `compute_mallows` always returns posterior distributions of the latent
 #'   consensus ranking \eqn{\rho} and the scale parameter \eqn{\alpha}. Several
@@ -44,9 +44,8 @@
 #'   Ulam distances when the cardinalities are not available, cf.
 #'   [get_cardinalities()].
 #'
-#' @param verbose Logical specifying whether to print out the progress of the
-#'   Metropolis-Hastings algorithm. If `TRUE`, a notification is printed
-#'   every 1000th iteration. Defaults to `FALSE`.
+#' @param progress_report An object of class "BayesMallowsProgressReported"
+#'   returned from [set_progress_report()].
 #'
 #' @param cl Optional cluster returned from [parallel::makeCluster()]. If
 #'   provided, chains will be run in parallel, one on each node of `cl`.
@@ -70,7 +69,7 @@ compute_mallows <- function(
     priors = set_priors(),
     initial_values = set_initial_values(),
     pfun_estimate = NULL,
-    verbose = FALSE,
+    progress_report = set_progress_report(),
     cl = NULL) {
   validate_class(data, "BayesMallowsData")
   validate_class(model_options, "BayesMallowsModelOptions")
@@ -92,7 +91,7 @@ compute_mallows <- function(
   } else {
     lapplyfun <- prepare_cluster(cl, c(
       "data", "model_options", "compute_options", "priors", "initial_values",
-      "pfun_values", "pfun_estimate", "verbose"
+      "pfun_values", "pfun_estimate", "progress_report"
     ))
     chain_seq <- seq_along(cl)
   }
@@ -109,7 +108,7 @@ compute_mallows <- function(
       initial_values = initial_values,
       pfun_values = pfun_values,
       pfun_estimate = pfun_estimate,
-      verbose = verbose
+      progress_report = progress_report
     )
   })
 

R/compute_mallows_mixtures.R

@@ -31,7 +31,7 @@ compute_mallows_mixtures <- function(
     priors = set_priors(),
     initial_values = set_initial_values(),
     pfun_estimate = NULL,
-    verbose = FALSE,
+    progress_report = set_progress_report(),
     cl = NULL) {
   stopifnot(is.null(cl) || inherits(cl, "cluster"))
 
@@ -40,15 +40,16 @@ compute_mallows_mixtures <- function(
   } else {
     lapplyfun <- prepare_cluster(cl, c(
       "data", "model_options", "compute_options", "priors", "initial_values",
-      "pfun_estimate", "verbose"
+      "pfun_estimate", "progress_report"
     ))
   }
 
   models <- lapplyfun(n_clusters, function(x) {
     model_options$n_clusters <- x
     compute_mallows(
-      data = data, model_options = model_options, compute_options = compute_options,
-      priors = priors, initial_values = initial_values, verbose = verbose
+      data = data, model_options = model_options,
+      compute_options = compute_options, priors = priors,
+      initial_values = initial_values, progress_report = progress_report
     )
   })
 

R/compute_mallows_sequentially.R

@@ -54,6 +54,21 @@ compute_mallows_sequentially <- function(
   if (!is.list(data) | !all(vapply(data, inherits, logical(1), "BayesMallowsData"))) {
     stop("data must be a list of BayesMallowsData objects.")
   }
+  if (any(
+    vapply(data, function(x) {
+      is.null(x$user_ids) || length(x$user_ids) == 0
+    }, logical(1))
+  )) {
+    stop("User IDs must be set.")
+  }
+  data <- lapply(data, function(x) {
+    if (!is.null(x$preferences)) {
+      x$preferences <- as.matrix(x$preferences)
+    } else {
+      x$preferences <- matrix(0, 0, 0)
+    }
+    x
+  })
   pfun_values <- extract_pfun_values(model_options$metric, data[[1]]$n_items, pfun_estimate)
   alpha_init <- sample(initial_values$alpha, smc_options$n_particles, replace = TRUE)
   rho_init <- initial_values$rho[, sample(ncol(initial_values$rho), smc_options$n_particles, replace = TRUE)]

R/estimate_partition_function.R

@@ -10,7 +10,7 @@
 #' any number of items, for footrule distances with up to 50 items, Spearman
 #' distance with up to 20 items, and Ulam distance with up to 60 items. This
 #' function is thus intended for the complement of these cases. See
-#' [get_cardinalities()] for details.
+#' [get_cardinalities()] and [compute_exact_partition_function()] for details.
 #'
 #' @param method Character string specifying the method to use in order to
 #'   estimate the logarithm of the partition function. Available options are
@@ -122,7 +122,7 @@ extract_pfun_values <- function(metric, n_items, pfun_estimate) {
           "estimate in argument pfun_estimate."
         )
       } else {
-        return(NULL)
+        NULL
       }
     }
   )