get_diag {stminsights}

computes stm model diagnostics

Description

get_diag() is a helper function to compute average and median semanticCoherence and exclusivity for a number of stm models. The function does not work for models with content covariates.

Usage

get_diag(models, outobj)

Arguments

models

A list of stm models.

outobj

The out object containing documents for all stm models.

Value

Returns model diagnostics in a data frame.

Examples

library(stm)
library(dplyr)
library(ggplot2)
library(quanteda)

# prepare data
data <- corpus(gadarian, text_field = 'open.ended.response')
docvars(data)$text <- as.character(data)

data <- tokens(data, remove_punct = TRUE) |>
  tokens_wordstem() |>
  tokens_remove(stopwords('english')) |> dfm() |>
  dfm_trim(min_termfreq = 2)

out <- convert(data, to = 'stm')

# fit models
gadarian_3 <- stm(documents = out$documents,
                  vocab = out$vocab,
                  data = out$meta,
                  prevalence = ~ treatment + s(pid_rep),
                  K = 3,
                  max.em.its = 1, # reduce computation time for example
                  verbose = FALSE)

gadarian_5 <- stm(documents = out$documents,
                  vocab = out$vocab,
                  data = out$meta,
                  prevalence = ~ treatment + s(pid_rep),
                  K = 5,
                  max.em.its = 1, # reduce computation time for example
                  verbose = FALSE)

# get diagnostics
diag <- get_diag(models = list(
                 model_3 = gadarian_3,
                 model_5 = gadarian_5),
                 outobj = out)
## Not run: 
# plot diagnostics
diag |>
ggplot(aes(x = coherence, y = exclusivity, color = statistic))  +
  geom_text(aes(label = name), nudge_x = 5) + geom_point() +
  labs(x = 'Semantic Coherence', y = 'Exclusivity') + theme_light()

## End(Not run)