# STA 437/1005 Assignment 1, Question 4.

# Read the data into a data frame.

d <- read.table("smog.txt",head=T)

# Put plots in a PDF file (one per page).

pdf("ass1-plots.pdf",pointsize=9)

# Scatterplots against year and against month.

par(mfrow=c(4,4))
for (i in 4:10) 
{ boxplot(d[,i]~d$year,xlab="year",ylab=colnames(d)[i])
  boxplot(d[,i]~d$month,xlab="month",ylab=colnames(d)[i])
}

# Produce all pairs of scatterplots, except those involving the year/month/day.

plot(d[-(1:3)],pch=".")

# Histograms of variables other than year/month/day.

par(mfrow=c(3,3))
for (i in 4:10)
{ hist(d[,i],nclass=30,xlab=colnames(d)[i],main="Histogram")
}

# Normal QQ plots of variables other than year/month/day.

par(mfrow=c(3,3))
for (i in 4:10)
{ qqnorm(d[,i],xlab=colnames(d)[i],pch=".")
}

# Histograms and normal QQ plots for some transformed variables.

par(mfrow=c(3,3))

hist(sqrt(d$deaths),nclass=30,xlab="Square root of deaths",main="Histogram")
qqnorm(sqrt(d$deaths),xlab="Square root of deaths",pch=".")

hist(sqrt(d$so2),nclass=30,xlab="Square root of so2",main="Histogram")
qqnorm(sqrt(d$so2),xlab="Square root of so2",pch=".")

hist(sqrt(d$ozone),nclass=30,xlab="Square root of ozone",main="Histogram")
qqnorm(sqrt(d$ozone),xlab="Square root of ozone",pch=".")

hist(log(d$pm10),nclass=30,xlab="Log of pm10",main="Histogram")
qqnorm(log(d$pm10),xlab="Log of pm10",pch=".")

# Make new data frame with some variables transformed.

nd <- d
nd$deaths <- sqrt(d$deaths); colnames(nd)[4] <- "sqrt deaths"
nd$so2 <- sqrt(nd$so2);      colnames(nd)[8] <- "sqrt so2"
nd$ozone <- sqrt(nd$ozone);  colnames(nd)[9] <- "sqrt ozone"
nd$pm10 <- log(nd$pm10);     colnames(nd)[10] <- "log pm10"

# Produce all pairs of scatterplots, with some variables transformed, except 
# those involving the year/month/day.

plot(nd[-(1:3)],pch=".")

# Done with plotting.

dev.off()