quanteda-run-through.r

ms = readRDS("mission-statements.rds")
glimpse(ms)
mscorp = corpus(ms$mission)
metadoc(mscorp, "unitid") = ms$unitid
ndoc(mscorp)
sum(ntoken(mscorp))
hist(ntoken(mscorp))
kwic(mscorp, "education", window=3)
msdfm = dfm(mscorp,
            stem = T,
            remove = stopwords(),
            remove_punct = T,
            remove_symbols = T) %>% 
  dfm_trim(min_termfreq = 5, 
           max_docfreq = .6,
           docfreq_type = "prop") %>% 
  dfm_tfidf()
topfeatures(msdfm)
topfeats = names(topfeatures(msdfm, 50))
fcm(msdfm) %>% 
  fcm_select(pattern =topfeats) %>% 
  textplot_network()
mat = as.matrix(msdfm)
k5 = kmeans(mat, 5, nstart = 25)
k5$size
plot(silhouette(k5$cluster, dist=dist(mat)), col=1:5, border=NA)