?dist ## good for continuous distances
library(cluster)
?daisy ## good for continuous, categorical, mixed distances/dissimilarities

load("/u/kalisch/teaching/12/ams/vorlesung/v4.2/dataV4.rda")
## interval scaled
head(usAirPol)
dist1 <- daisy(usAirPol, metric = "euclidean")
dist1
mds1 <- cmdscale(dist1)
plot(mds1, asp = 1)
dist2 <- daisy(usAirPol, metric = "euclidean", stand = TRUE)
mds2 <- cmdscale(dist2)
plot(mds2, asp = 1)
identify(mds2, labels = rownames(usAirPol))

## symm. bin.
## Which persons are similar (measured by sex/married/fair hair/glasses)?
str(dfSymBin)
dfSymBin
dist3 <- daisy(dfSymBin, type = list(symm = 1:4))
dist3
dfSymBin
library(MASS)
mds3 <- isoMDS(dist3)
plot(mds3$points, asp = 1, type = "n")
text(mds3$points, labels = rownames(dfSymBin))
dfSymBin

## asym.bin.
## Which documents are similar (measures by appearance of the words four/referee/goal/return/stock/price)
str(dfAsymBin)
dist4 <- daisy(dfAsymBin, type = list(asymm = 1:6))
dist4
dfAsymBin
mds4 <- isoMDS(dist4)
plot(mds4$points, asp = 1, type = "n")
text(mds4$points, labels = rownames(dfAsymBin))
dfAsymBin
## generating data done

## Nominal
## what do you like most out of a selection of things?
str(dfNom)
dist5 <- daisy(dfNom)
dist5
mds5 <- isoMDS(dist5)
plot(mds5$points, asp = 1, type = "n")
text(mds5$points, labels = rownames(dfNom))
dfNom

## Ordinal
## How much do you agree on (0-10) on 5 questions?
datOrd
dfOrd <- as.data.frame(datOrd)
for (i in 1:5) dfOrd[[i]] <- ordered(dfOrd[[i]])
str(dfOrd)

dist6 <- daisy(dfOrd)
dist6
mds6 <- isoMDS(dist6)
plot(mds6$points, asp = 1, type = "n")
text(mds6$points, labels = rownames(dfOrd))
datOrd

# Mixed
## flowers
data(flower)
?flower
str(flower)

## use default data types for variables
d1 <- daisy(flower, metric = "gower", type = list(asymm = c(1,3), symm = 2))
d1 ## Types = A, S, A, N, O, O, I, I 
cmd1 <- cmdscale(d1, eig = TRUE)
cmd1
cmd2 <- isoMDS(d1)
plot(cmd2$points, type = "n")
text(cmd2$points, labels = rownames(flower))

flower[c(16,17),]
flower[c(5,17),]