# set working directory
setwd("/home/ekrem/Project2/Data")

# include the package
library(randomForest)

# define a variable for error rates
errorRates = matrix(0,1,500)

# load the dataset
mydata <- read.table("glass.csv",header=TRUE, sep=",")

# start fitting random forests
mydata.rf <- randomForest(x=mydata[,2:10], y=mydata[,11],importance=TRUE, proximity=TRUE, ntree=500, type="classification")
myErrRate = mydata.rf[["err.rate"]]

# below line is for the importance of variables
round(importance(mydata.rf), 2)

plot(myErrRate[,1],type="o",col='RED',xlab="Iteration",ylab="Error Rate",ylim=cbind(0,0.7),xlim=cbind(0,550))
#lines(myErrRate[,2],type="o",col='BLUE')
#lines(myErrRate[,3],type="o",col='BLUE')
#lines(myErrRate[,4],type="o",col='BLUE')
#lines(myErrRate[,5],type="o",col='BLUE')
#lines(myErrRate[,6],type="o",col='BLUE')
#lines(myErrRate[,7],type="o",col='BLUE')
#g_range <- range(0, myErrRate[,1], myErrRate[,2])
title(main="Random Forest Error Rates", col.main="red", font.main=16)
legend(1, g_range[2], c("Err. Rate"), cex=0.8,col=c("red"), pch=21:22, lty=1:2);



# start fitting random forests
mydata.rf1 <- randomForest(x=mydata[,2:10], y=mydata[,11],importance=TRUE, proximity=TRUE, ntree=200, type="classification",mtry=1)
myErrRate1 = mydata.rf1[["err.rate"]]

mydata.rf2 <- randomForest(x=mydata[,2:10], y=mydata[,11],importance=TRUE, proximity=TRUE, ntree=200, type="classification",mtry=2)
myErrRate2 = mydata.rf2[["err.rate"]]

mydata.rf3 <- randomForest(x=mydata[,2:10], y=mydata[,11],importance=TRUE, proximity=TRUE, ntree=200, type="classification",mtry=3)
myErrRate3 = mydata.rf3[["err.rate"]]

mydata.rf4 <- randomForest(x=mydata[,2:10], y=mydata[,11],importance=TRUE, proximity=TRUE, ntree=200, type="classification",mtry=4)
myErrRate4 = mydata.rf4[["err.rate"]]

mydata.rf5 <- randomForest(x=mydata[,2:10], y=mydata[,11],importance=TRUE, proximity=TRUE, ntree=200, type="classification",mtry=5)
myErrRate5 = mydata.rf5[["err.rate"]]

mydata.rf6 <- randomForest(x=mydata[,2:10], y=mydata[,11],importance=TRUE, proximity=TRUE, ntree=200, type="classification",mtry=6)
myErrRate6 = mydata.rf6[["err.rate"]]

mydata.rf7 <- randomForest(x=mydata[,2:10], y=mydata[,11],importance=TRUE, proximity=TRUE, ntree=200, type="classification",mtry=7)
myErrRate7 = mydata.rf7[["err.rate"]]

mydata.rf8 <- randomForest(x=mydata[,2:10], y=mydata[,11],importance=TRUE, proximity=TRUE, ntree=200, type="classification",mtry=8)
myErrRate8 = mydata.rf8[["err.rate"]]

mydata.rf9 <- randomForest(x=mydata[,2:10], y=mydata[,11],importance=TRUE, proximity=TRUE, ntree=200, type="classification",mtry=9)
myErrRate9 = mydata.rf9[["err.rate"]]

plot(myErrRate1[,1],type="o",col='RED',xlab="Iteration",ylab="Error Rate",ylim=cbind(0,0.5),xlim=cbind(0,250))
title(main="Error Rates for mtry = 1", col.main="red", font.main=16)

plot(myErrRate2[,1],type="o",col='RED',xlab="Iteration",ylab="Error Rate",ylim=cbind(0,0.5),xlim=cbind(0,250))
title(main="Error Rates for mtry = 2", col.main="red", font.main=16)

plot(myErrRate3[,1],type="o",col='RED',xlab="Iteration",ylab="Error Rate",ylim=cbind(0,0.5),xlim=cbind(0,250))
title(main="Error Rates for mtry = 3", col.main="red", font.main=16)

plot(myErrRate4[,1],type="o",col='RED',xlab="Iteration",ylab="Error Rate",ylim=cbind(0,0.5),xlim=cbind(0,250))
title(main="Error Rates for mtry = 4", col.main="red", font.main=16)

plot(myErrRate5[,1],type="o",col='RED',xlab="Iteration",ylab="Error Rate",ylim=cbind(0,0.5),xlim=cbind(0,250))
title(main="Error Rates for mtry = 5", col.main="red", font.main=16)

plot(myErrRate6[,1],type="o",col='RED',xlab="Iteration",ylab="Error Rate",ylim=cbind(0,0.5),xlim=cbind(0,250))
title(main="Error Rates for mtry = 6", col.main="red", font.main=16)

plot(myErrRate7[,1],type="o",col='RED',xlab="Iteration",ylab="Error Rate",ylim=cbind(0,0.5),xlim=cbind(0,250))
title(main="Error Rates for mtry = 7", col.main="red", font.main=16)

plot(myErrRate8[,1],type="o",col='RED',xlab="Iteration",ylab="Error Rate",ylim=cbind(0,0.5),xlim=cbind(0,250))
title(main="Error Rates for mtry = 8", col.main="red", font.main=16)

plot(myErrRate9[,1],type="o",col='RED',xlab="Iteration",ylab="Error Rate",ylim=cbind(0,0.5),xlim=cbind(0,250))
title(main="Error Rates for mtry = 9", col.main="red", font.main=16)