#
auto <- read.table("../data/auto.dat", header=TRUE)

set.seed(34534)
library(tree)
n = nrow(auto)
train = sample(1:n,101,replace=FALSE)

t1 <- tree(city.distance~engine.size+curb.weight,
           data=auto[train,])
pred1 = predict(t1,auto[-train,])
show(sum((auto$city.distance[-train]-pred1)^2))
t2 <- tree(city.distance~engine.size+curb.weight,data=auto[-train,])

par(mfrow=c(1,2))
plot(t1)
text(t1, digits=2, pretty=3,cex=0.5)
plot(t2)
text(t2, digits=2, pretty=3,cex=0.5)

p1<- prune.tree(t1, newdata=auto[-train,])

p1<- prune.tree(t1, newdata=auto[-train,],best=6)
plot(p1)
text(p1, digits=2, pretty=3,cex=0.5)
pred2 = predict(p1,auto[-train,])
show(sum((auto$city.distance[-train]-pred2)^2))

p2<- prune.tree(t2, newdata=auto[train,],best=5)

par(mfrow=c(1,2))
plot(p1)
text(p1, digits=2, pretty=3)
plot(p2)
text(p2, digits=2, pretty=3)