## Cluster Analysis: DISTANZE E DISSIMILARITA'

library("ggplot2")
library("cluster")

# food_spending.csv
FoodS <- read.csv("food_spending.csv", header = TRUE)
View(FoodS)  
dim(FoodS)
#scatterplot
ggplot(FoodS, aes(FreqAway, Home))+
  geom_point()
str(FoodS)
#plot(FoodS$FreqAway, FoodS$Home)

# Distanze/dissimilarità
?dist

Dist.e<-dist(FoodS)
Dist.e
str(Dist.e)
head(Dist.e)

Dmat<-as.matrix(Dist.e)
head(Dmat)
dim(Dmat)
Dmat[1:5, 1:5]

#install.packages("cluster")
#library("cluster")

?daisy
dist1<-daisy(FoodS)  # Euclidean dist.
as.matrix(dist1)[1:5, 1:5]

# standardization
dist2<-daisy(FoodS, stand = TRUE)
# as.matrix..

### Esercizio: distanza di Manhattan su dati standardizzati
s.FoodS<-scale(FoodS) # scaled matrix
?dist
Md2<-as.matrix(dist(scale(FoodS), method = "manhattan"))
hist(Md2)

## 28/04/25

library("cluster")
d<-read.csv(file="claims.csv", header = TRUE)
head(d)
summary(d)
# litig: 0=no, 1=si
# soft_injury: 0=no, 1=si
# emergency_tr: 0=no, 1=si
# NumTreat: numero di cure mediche 

d$litig<-as.factor(d$litig)
d$soft_injury<-as.factor(d$soft_injury)
d$emergency_tr<-as.factor(d$emergency_tr)

#save(d, file = "claims2.RData")

load("claims2.RData")

table(d$litig)
prop.table(table(d$litig))
barplot(prop.table(table(d$litig)), col=2)
# levels()

# compute Gower's dissimilarity

diss_g<-as.matrix(daisy(d, metric = "gower"))
diss_g[1:5, 1:5]
range(diss_g)
which(diss_g==max(diss_g), arr.ind = TRUE)[1,]
d[656, ]
d[69, ]