Assignment probability heatmap (BT-SBM)

Heatmap of item-wise posterior assignment probabilities for clusters (relabeled so that Cluster 1 is the top block by decreasing \(\lambda\)). Items are ordered by their most probable cluster and marginal wins.

Usage

plot_assignment_probabilities(
  fit,
  w_ij = NULL,
  max_n_clust = NULL,
  clean_fun = clean_players_names,
  k_show = NULL,
  fill_low = "#FFFFCC",
  fill_high = "#006400"
)

Arguments

fit

Output list from gibbs_BT_SBM() (must include relabeled$assign_prob).

w_ij

Optional wins matrix to compute marginal wins for ordering and annotation. If NULL, items are ordered by most-probable cluster only.

max_n_clust

Where to filter the mcmc x_t. If not specified we use the modal K

clean_fun

Optional function to prettify names. Default: identity.

k_show

Optional integer number of clusters to show (defaults to all columns in assign_prob).

fill_low, fill_high

Colors for the heatmap gradient low/high.

Value

A ggplot object.

Examples

if (FALSE) { # \dontrun{
p <- plot_assignment_probabilities(fit, w_ij)
} # }