---
title: "faktorid2"
output: html_document
---

```{r}
 library(languageR)
 head(dutchSpeakersDist)
 head(dutchSpeakersDistMeta)
 dutchSpeakersDist.d=dist(dutchSpeakersDist) 
 dutchSpeakersDist.mds=cmdscale(dutchSpeakersDist.d, k=3) #ruumiline
head(dutchSpeakersDist.mds)
p=dutchSpeakersDist.mds
 #install.packages("plot3D")
library(plot3D)
points3D(p[,1], p[,2], p[,3])
 
asukohad=cmdscale(dutchSpeakersDist.d, k=2) #tasand
head(asukohad)
plot(asukohad) 
 
 
asukohad=cmdscale(dutchSpeakersDist.d, k=1) # joon
head(asukohad)
plot(asukohad, sample(0, nrow(asukohad), replace = TRUE), col=rgb(0, 0, 0, 0.3)) 


#Paigutage multidimensionaalse skaleerimise abil kahemõõtmelisele joonisele 
#ühe päeva tunnid, nii et sarnasema ilmaga tunnid oleksid lähestikku


ilm=read.table("http://www.tlu.ee/~jaagup/andmed/ilm/harkutund.txt", header=TRUE, sep=",")
ilmtervik=ilm[complete.cases(ilm), ]
ilmskaleeritud=scale(ilmtervik[1:24, 5:11])
ilmskaleeritud
head(ilm, 24)
asukohad=cmdscale(dist(ilmskaleeritud), 2)
asukohad
plot(asukohad, type="n")
text(asukohad, cex=0.5)

plot(cmdscale(dist(scale(ilmtervik[1:2000, 4:11])), 2))

asukohad=cmdscale(dist(scale(ilmtervik[1:2000, 4:11])), 2)
plot(asukohad, type="n")
text(x=asukohad[, 1], y=asukohad[, 2], labels=ilmtervik[1:2000, "Kuu"])

ryhmad=kmeans(asukohad, centers=3)
plot(asukohad, type="n")
text(x=asukohad[, 1], y=asukohad[, 2], labels=ryhmad$cluster)

plot(asukohad, col=ryhmad$cluster)

head(lexicalMeasures)

plot(hclust(dist(cor(lexicalMeasures[, -1]))))

ilm=read.table("http://www.tlu.ee/~jaagup/andmed/ilm/harkutund.txt", header=TRUE, sep=",")
ilmtervik=ilm[complete.cases(ilm), ]
ilmskaleeritud=scale(ilmtervik[2000:5000, 4:11])
plot(hclust(dist(cor(ilmskaleeritud))))

jaanuar1=ilmtervik[1:24, 5:11]
plot(hclust(dist(cor(scale(jaanuar1)))))
plot(hclust(dist(cor(scale(t(jaanuar1))))))
jaanuar1
t(jaanuar1)

ilmskaleeritud=scale(ilmtervik[ilmtervik$Kuu==1 & ilmtervik$Kell=="10:00", 4:11])
plot(hclust(dist(cor(ilmskaleeritud))))
row.names(ilmskaleeritud)=1:31
plot(hclust(dist(cor(t(ilmskaleeritud)))))

ryhmad=cutree(hclust(dist(cor(t(ilmskaleeritud)))), k=3) #joon kolme haru kohalt
plot(ilmskaleeritud[, c("TA1H", "WS1H")], type="n") #teljed
text(ilmskaleeritud[, c("TA1H", "WS1H")], 
     labels = row.names(ilmskaleeritud),  col=ryhmad)


```