R17

			Random Forest Classifiers


# Heart Attack Data

# Classification Tree (from R16)

library(rpart)

heart <- read.table("F:/S5600/Data sets/SAheart.txt")
head(heart)

pairs(heart, col=heart$chd+1)

heart.cart<-rpart(chd~., data=heart, method="class")
heart.cart

plot(heart.cart,margin=.2)
text(heart.cart, use.n=FALSE, pretty=0, cex=.8)

heart.cart.pred <- predict(heart.cart, heart[,-10], type="class")
table(Actual=heart$chd, Classified=heart.cart.pred)

nrow(heart)
(27+71)/462	# Overall Error Rate

# Bootstrap sample

boot.samp <- sample(462,462,replace=T)
sort(boot.samp)[1:50]

heart.boot <- heart[boot.samp,]

heart.b.cart <- rpart(chd~., data=heart.boot, method="class")

plot(heart.b.cart)
text(heart.b.cart, use.n=FALSE, pretty=0, cex=.8)

# Repeat bootstrap tree - note differences

# Random Forests

library(randomForest)	# Must be installed

heart$chd <- factor(heart$chd, labels=c("absent","present")) 
# Response MUST be a factor for classification random forests


heart.rf <- randomForest(chd~., data=heart, importance=T,
	proximity = T)

heart.rf
plot(heart.rf)	# plot cumulative error rates - black (overall),
	# red - class 0 (absent), green - class 1 (present)

# variable importance

importance(heart.rf)
varImpPlot(heart.rf)

pairs(heart[,c(2,3,4,5,9)],pch=unclass(heart$chd),col=unclass(heart$chd))

# MDS using RF proximities

MDSplot(heart.rf, heart$chd, pch=unclass(heart$chd))

# Sample more balanced sets - balance error rates?

heart.rf.bal <- randomForest(chd~., data=heart, sampsize=c(100,100),
		importance=T, proximity = T)
heart.rf.bal
plot(heart.rf.bal)

importance(heart.rf.bal)
varImpPlot(heart.rf.bal)

MDSplot(heart.rf.bal, heart$chd, pch=unclass(heart$chd))

# Crabs Data

library(MASS)
data(crabs)

crabs <- crabs[,c(1,4:8)]

crabs.rf <- randomForest(sp~., data=crabs, importance=T,
	proximity = T)
crabs.rf

plot(crabs.rf)

importance(crabs.rf)
varImpPlot(crabs.rf)


# Forensic Glass Data

library(MASS)
data(fgl)

head(fgl)

fgl.rf <- randomForest(type~., data=fgl, importance=T,
	proximity = T)
fgl.rf

plot(fgl.rf)

importance(fgl.rf)
varImpPlot(fgl.rf)