Skip to contents

Find Optimal Number of Topics

Source: R/topic_modeling.R
find_optimal_k.Rd

Searches for the optimal number of topics (K) using stm::searchK. Produces diagnostic plots to help select the best K value.

Usage

find_optimal_k(
  dfm_object,
  topic_range,
  max.em.its = 75,
  emtol = 1e-04,
  cores = 1,
  categorical_var = NULL,
  continuous_var = NULL,
  height = 600,
  width = 800,
  verbose = TRUE,
  ...
)

Arguments

dfm_object

A quanteda dfm object to be used for topic modeling.

topic_range

A vector of K values to test (e.g., 2:10).

max.em.its

Maximum number of EM iterations (default: 75).

emtol

Convergence tolerance for EM algorithm (default: 1e-04). Higher values (e.g., 1e-03) speed up fitting but may reduce precision.

cores

Number of CPU cores to use for parallel processing (default: 1). Set to higher values for faster searchK on multi-core systems.

categorical_var

Optional categorical variable(s) for prevalence.

continuous_var

Optional continuous variable(s) for prevalence.

height

Plot height in pixels (default: 600).

width

Plot width in pixels (default: 800).

verbose

Logical indicating whether to print progress (default: TRUE).

...

Additional arguments passed to stm::searchK.

Value

A list containing search results and diagnostic plots.

See also

Other topic-modeling: analyze_semantic_evolution(), assess_embedding_stability(), assess_hybrid_stability(), auto_tune_embedding_topics(), calculate_assignment_consistency(), calculate_eval_metrics_internal(), calculate_keyword_stability(), calculate_semantic_drift(), calculate_topic_probability(), calculate_topic_stability(), find_topic_matches(), fit_embedding_model(), fit_hybrid_model(), fit_temporal_model(), generate_topic_labels(), get_topic_prevalence(), get_topic_terms(), get_topic_texts(), identify_topic_trends(), plot_model_comparison(), plot_quality_metrics(), run_contrastive_topics_internal(), run_neural_topics_internal(), run_temporal_topics_internal(), validate_semantic_coherence()