Package index • tmfast

Generators

Generate a simulated text corpus

peak_alpha(): Alpha parameter with a single peak

expected_entropy(): Expected entropy for samples from a Dirichlet distribution

rdirichlet(): Sample from the Dirichlet distribution

draw_corpus(): Draw a collection of documents

journal_specific(): "Journal-specific" simulation scenario

Information gain

Tools for information-theoretic vocabulary selection

ndH(): Information gain (uniform distribution)

ndR(): Information gain (length-proportional distribution)

tmfast

Fitting topic models with PCA+varimax

tmfast(): Fit a topic model using PCA+varimax

insert_topics(): Insert a topic model into a fitted tmfast

varimax_irlba(): Fit a varimax-rotated PCA using irlba

fit_varimax(): Given a (rank n) PCA fit, return a rank k < n varimax fit

predict(<varimaxes>): Project new data into PCA score space

Tidiers

Extract beta and gamma matrices from tmfast objects

tidy(<tmfast>): Extract beta and gamma matrices from tmfast objects

tidy_all(): Extract gamma or beta matrices for all topics

Renormalization

Renormalize a distribution to match a desired expected entropy

solve_power(): Solve the equation to find the desired exponent

target_power(): Find target power for renormalization

renorm(): Renormalize tidied distributions

Hellinger distances

Calculate Hellinger distances between distributions

hellinger(): Hellinger distances

compare_betas(): Compare topic-word distributions using Hellinger distance

Discursive space visualizations

tsne(): Discursive space using t-SNE

umap(): Discursive space using UMAP

Utilities

Utility functions

build_matrix(): Convert a long dataframe to a wide (sparse) matrix

entropy(): Entropy of a distribution

loadings(): Extract a PCA/varimax loadings matrix

scores(): Extract item scores from a fitted PCA/varimax model

rotation(): Extract varimax rotation

Package

tmfast-package: Fitting "topic models" with PCA+varimax