predict_top_k: Predict Top-k Rankings with Pairwise Preferences

In BayesMallows: Bayesian Preference Learning with the Mallows Rank Model

Predict Top-k Rankings with Pairwise Preferences

Description

Predict the posterior probability, per item, of being ranked among the top-k for each assessor. This is useful when the data take the form of pairwise preferences.

Usage

predict_top_k(model_fit, burnin = model_fit$burnin, k = 3)

Arguments

model_fit	An object of type BayesMallows, returned from compute_mallows.
burnin	A numeric value specifying the number of iterations to discard as burn-in. Defaults to model_fit$burnin, and must be provided if model_fit$burnin does not exist. See assess_convergence.
k	Integer specifying the k in top-k.

Value

A dataframe with columns assessor, item, and prob, where each row states the probability that the given assessor rates the given item among top-k.

See Also

plot_top_k

Other posterior quantities: assign_cluster(), compute_consensus.BayesMallows(), compute_consensus.SMCMallows(), compute_consensus(), compute_posterior_intervals.BayesMallows(), compute_posterior_intervals.SMCMallows(), compute_posterior_intervals(), heat_plot(), plot.BayesMallows(), plot.SMCMallows(), plot_elbow(), plot_top_k(), print.BayesMallowsMixtures(), print.BayesMallows()

Examples

## Not run: 
  # We use the example dataset with beach preferences. Se the documentation to
  # compute_mallows for how to assess the convergence of the algorithm
  # We need to save the augmented data, so setting this option to TRUE
  model_fit <- compute_mallows(preferences = beach_preferences,
                               save_aug = TRUE)
  # We set burnin = 1000
  model_fit$burnin <- 1000
  # By default, the probability of being top-3 is plotted
  plot_top_k(model_fit)
  # We can also plot the probability of being top-5, for each item
  plot_top_k(model_fit, k = 5)
  # We get the underlying numbers with predict_top_k
  probs <- predict_top_k(model_fit)
  # To find all items ranked top-3 by assessors 1-3 with probability more than 80 %,
  # we do
  subset(probs, assessor %in% 1:3 & prob > 0.8)


## End(Not run)

BayesMallows documentation built on Nov. 25, 2023, 5:09 p.m.