####################################################################################
#This is a file with executable R code for chapter 16 of Natalia Levshina's (2015) 
#How to Do Linguistics with R. Amsterdam/Philadelphia: John Benjamins. 
####################################################################################


###Section 16.2

##Main text

install.packages("cluster")

library(Rling); library(cluster)

data(cooking)

head(cooking)

cooking <- cooking[rowSums(cooking) > 0, ]

exp.bake <- sum(cooking$Bake)*rowSums(cooking)/sum(cooking)

exp.bake[1:5]

PMI.bake <- log2(cooking$Bake/exp.bake)

PMI.bake[1:5]

PPMI.bake <- ifelse(PMI.bake < 0, 0, PMI.bake)

PPMI.bake[1:5]

cooking <- as.matrix(cooking)

cooking.exp <- chisq.test(cooking)$expected

cooking.PMI <- log2(cooking/cooking.exp)

cooking.PPMI <- ifelse(cooking.PMI < 0, 0, cooking.PMI)

crossprod(cooking.PPMI[, 1], cooking.PPMI[, 10])/sqrt(crossprod(cooking.PPMI[, 1]) *crossprod(cooking.PPMI[, 10]))

crossprod(cooking.PPMI[, 2], cooking.PPMI[, 10])/sqrt(crossprod(cooking.PPMI[, 2]) *crossprod(cooking.PPMI[, 10]))

cooking1 <- t(cooking.PPMI)

cooking.cos <- cossim(cooking1)

round(cooking.cos, 2)

cooking.dist <- 1 - (cooking.cos/0.07958)

cooking.dist <- as.dist(cooking.dist)

test.clust <- pam(cooking.dist, 2)

test.clust$silinfo$avg.width

test.clust <- pam(cooking.dist, 6)

test.clust$clustering