# select top 100 words by avg rank
hn_counts_and_ranks %>% top_n(100, wt = avg_rank) %>%

# construct plot
ggplot(aes(avg_rank, n, label = word)) +
  geom_text_repel(segment.alpha = 0, aes(colour=avg_rank,
                                         size=n)) +
  scale_color_gradient(low="green3", high="violetred", trans = "log10",
                       guide = guide_colourbar(direction = "horizontal",
                                               title.position ="top")) +
  scale_size_continuous(range = c(3, 10),
                        guide = FALSE) +
  scale_x_log10() +
  ggtitle(paste0("Top 100 words from ",
                  nrow(hn_comments),
                 " Hacker News article comments, by average ranking"),
          subtitle = "word frequency (size) ~ avg comment ranking (color)") +
  labs(y = "Word frequency", x = "Avg rank (log scale)") +
  theme_minimal() +
  theme(legend.position=c(.99, .99),
        legend.justification = c("right","top"),
        panel.grid.major = element_line(colour = "whitesmoke"))