plot_mi_check: Visualize a topic's MI score in comparison to simulated...

In agoldst/dfrtopics: Tools for exploring topic models of text

Visualize a topic's MI score in comparison to simulated values

Description

As a diagnostic visualization, this function displays the MI score of a topic in comparison with the distribution of simulated values, on a scale normalized to the latter. Extreme deviations of the actual MI score indicates a departure from the multinomial assumption of the model. Exploring these deviations over various document groupings may help to reveal the driving factor of those deviations (e.g. variation over time).

Usage

plot_mi_check(m, k, groups = NULL, n_reps = 20)

Arguments

m	mallet_model object with sampling state loaded via load_sampling_state
k	topic number (only one topic at a time)
groups	optional grouping factor with one element for each document
n_reps	number of simulations

Value

ggplot2 plot object

See Also

mi_topic, mi_check, plot_imi_check

agoldst/dfrtopics documentation built on July 15, 2022, 4:13 p.m.