publish-date-ungram-tfidf.R

plot_ungram_tfidf = function(dfin, start, stend, slice_unit="year", nmin=2, ntop=10, ncol=3) {
# ---- publish-date-ungram-tfidf
ungram_tf_idf <- dfin %>%
 subset(select=c(ungram,publish_date)) %>%
 mutate(Date = as.Date(substr(publish_date,1,8), format="%Y%m%d")) %>%
 filter(Date>=start & Date<=stend) %>%
 mutate(time_slice = floor_date(Date, unit = slice_unit)) %>%
 mutate(time_slice = format(time_slice, "%Y-%m-%d")) %>%
 count(time_slice, ungram) %>%
 filter(between(n,nmin,max(n))) %>%
 bind_tf_idf(ungram, time_slice, n) %>%
 arrange(desc(tf_idf))
ungram_tf_idf.plot <- ungram_tf_idf %>%
 mutate(ungram = factor(ungram, levels = rev(unique(ungram)))) %>%
 arrange(time_slice, tf_idf) %>%
 group_by(time_slice) %>% 
 top_n(n = ntop) %>% 
 ungroup()
plot_title <- paste("Ntop", ntop, "ungram tf-idf scores", start, "to", stend, "slby", slice_unit)
ungram_tf_idf.plot %>%
 ggplot(aes(x = ungram, y = tf_idf, fill = time_slice)) +
 geom_col(show.legend = FALSE) +
 labs(x = "un.grm", y = "tf-idf") +
 facet_wrap(~time_slice, ncol = ncol, scales = "free") +
 theme(text = element_text(size = 10)) +
 ggtitle(plot_title) +
 coord_flip()
}