Visualize alignments between topics of distinct LDA models

This function creates flow diagrams of alignment weight across collections of topics. It is the default plot method associated with alignment objects. The x-axis indexes models from those with few to those with many topics. Each rectangle matches a topic; its height is the weight associated with that topic from across all samples. The width of edges between topics corresponds of the alignment weight between that pair of topics. By default, topics are shaded according to their path. Other topic measures can be provided by modifying the `color_by` argument.

plot_alignment(
  x,
  rect_gap = 0.2,
  color_by = "path",
  model_name_repair_fun = paste0,
  label_topics = FALSE,
  add_leaves = FALSE,
  leaves_text_size = 10,
  n_features_in_leaves = 3,
  min_feature_prop = 0.1,
  top_n_edges = NULL
)

Arguments

x

(required) An alignment class object resulting from align_topics.

rect_gap

(optional) A float describing how much vertical space to put between topics within the same model. The units correspond to topic masses. Defaults to 0.2.

color_by

(optional) What should the color of topics and weights encode? Defaults to 'path'. Other possible arguments are 'coherence', 'refinement', or 'topic'.

model_name_repair_fun

(optional) How should names be repaired before plotting?

label_topics

(optional, default = FALSE) A logical specifying if topics should be labeled with the "color_by" information.

add_leaves

(optional, default = FALSE) A logical specifying if the topic composition of leave-topics should be printed.

leaves_text_size

(optional, default = 10) specifies the font size of leaves annotations in pt if add_leaves is TRUE.

n_features_in_leaves

(optional, default = 3) specifies the maximum number of features that should be included in the leaves annotations if add_leaves is TRUE.

min_feature_prop

(optional, default = 0.1) specifies the minimum proportion of a feature in a topic for that feature to be included in the leaves annotations if add_leaves is TRUE.

top_n_edges

(optional, integer, default = NULL) specifies the number of edges that should be drawn between the topics of subsequent models. The top_n_edges with the highest weights are drawn. If NULL (default), all edges are drawn.

Value

A ggplot2 object describing the alignment weights across models.

See also

align_topics