Struct CorrelatedTopicModel 

pub struct CorrelatedTopicModel {
    pub config: CtmConfig,
    /* private fields */
}

Correlated Topic Model estimator.

Fit with CorrelatedTopicModel::fit; the fitted result is returned as a CtmResult value (the model itself is stateless after construction).

Fields

config: CtmConfig

Model configuration.

Implementations

impl CorrelatedTopicModel

pub fn fit( &self, doc_counts_list: &[Vec<f64>], vocab_size: usize, ) -> Result<CtmResult>

Fit the CTM to a collection of documents represented as word-count vectors.

Arguments

• doc_counts_list – one count vector per document (length V each)
• vocab_size – vocabulary size V (must equal doc_counts_list[d].len())

Returns

A CtmResult containing the fitted parameters.

impl CorrelatedTopicModel

pub fn new(config: CtmConfig) -> Self

Construct a new (unfitted) CTM with the given configuration.

pub fn fitted_result(&self) -> Option<&CtmResult>

Return a reference to the fitted result, if available.

pub fn fit_and_store( &mut self, doc_counts_list: &[Vec<f64>], vocab_size: usize, ) -> Result<&CtmResult>

Fit the model and store the result internally, also returning it.

pub fn top_words_from_fitted( &self, vocab: &[String], n: usize, ) -> Option<Vec<Vec<String>>>

Return the top-n words for each topic given a vocabulary.

Requires the model to have been fitted (via fit_and_store).

pub fn correlation_matrix_from_fitted(&self) -> Option<Vec<Vec<f64>>>

Compute the inter-topic correlation matrix from the fitted Σ.

Trait Implementations

impl Default for CorrelatedTopicModel

fn default() -> Self

Returns the “default value” for a type.