Simple frequency plot of the bigram

bigram.R

library(dplyr)
library(ggplot2)
#see following
#browseURL("https://rekihaku.pref.hyogo.lg.jp/curator/20165/")

wazauta = scan("wazauta.txt", what = character())
n = nchar(wazauta)
unigram = character(n)
for( i in 1:n ){
  unigram[i] = substr(wazauta, i, i)
}

nline = 20L
df_bi = data.frame(unigram=unigram) %>% 
  mutate(pos=row_number(), nextchar=lead(unigram)) %>% 
  group_by(unigram, nextchar) %>% 
  mutate(freq=n()) %>% 
  ungroup() %>% 
  mutate(x = pos %/% nline, y = pos %% nline)
head(df_bi)

p = ggplot(df_bi, aes(x=x, y=y))+
  geom_text(aes(label = unigram, colour=freq), family="Osaka")+
  scale_color_continuous()+
  scale_y_reverse()+scale_x_reverse()+
  theme_minimal(20)
print(p)
ggsave(plot=p, filename = "wazauta.png", width = 9, height = 9)