Residual dispersion check (is K large enough?)

Multinomial residual dispersion (Taddy 2012; port of stm::checkResiduals). A dispersion well above 1 suggests too few topics.

Usage

check_residuals(model, tol = 0.01)

Arguments

model

A faSTM fit (carries its documents).

tol

Threshold for counting estimable residual cells.

Value

A list with dispersion, pvalue, and df.