ilm=read.table("http://www.tlu.ee/~jaagup/andmed/ilm/harkutund.txt", header=TRUE, sep=",", )
temperatuurid=ilm[!is.na(ilm[, "TA1H"]), "TA1H"]
vastus1=kmeans(temperatuurid, centers=4, nstart=20)
vastus1$centers
vastus1$withinss
vastus1$totss
vastus1$betweenss
vastus1$tot.withinss
plot(temperatuurid, col=vastus1$cluster)

plot(2:8, sapply(2:8, function(n){
  v=kmeans(temperatuurid, centers=n, nstart=20)
  v$tot.withinss
}), ylab="pyydmata erinevused")


#Klasterdage joonisel temperatuuri/
#tuule tugevuse (WS1h) tasandil n klastriks

plot(ilm[, c("TA1H", "WS1H")])
ilm2=ilm[complete.cases(ilm), c("TA1H", "WS1H")]
ilm3=scale(ilm2)
vastus2=kmeans(ilm3, centers=4)
plot(ilm2[, c("TA1H", "WS1H")], col=vastus2$cluster)


plot(ilm[1:24, "TA1H"])
dist(ilm[1:24, "TA1H"])
hclust(dist(ilm[1:24, "TA1H"]))
plot(hclust(dist(ilm[1:24, "TA1H"])))

plot(hclust(dist(ilm[1:24, "TA1H"]), method = "average"))
plot(hclust(dist(ilm[1:24, "TA1H"]), method = "single"))
plot(hclust(dist(ilm[1:24, "TA1H"]), method = "median"))


ilm2=ilm[1:24, c("TA1H", "WS1H")]
ilm3=scale(ilm2)
vastus2=kmeans(ilm3, centers=3)
plot(ilm2[, c("TA1H", "WS1H")], col=vastus2$cluster)
plot(hclust(dist(ilm3)))
m=cutree(hclust(dist(ilm3)), k=3)
plot(ilm2[, c("TA1H", "WS1H")], col=m)


pikkus=70:100
mass=jitter(pikkus*0.16, 50) 
plot(pikkus, mass)
lapsed=data.frame(pikkus, mass)
lapsed
summary(prcomp(lapsed, scale=TRUE))
prcomp(lapsed, scale=TRUE)
?jitter


prcomp(ilm[complete.cases(ilm), 4:11], scale=TRUE)
summary(prcomp(ilm[complete.cases(ilm), 4:11], scale=TRUE))