####################################################################

# [ reading in data ]

prestige <- read.table("prestige.dat")

prestige[1:10,]

names(prestige) _ c(
 "education",
 "income",
 "pct.female",
 "occ.prestige",
 "occ.code",
 "occ.type"
 )

prestige[1:10,]

###################################################################

# [ assignment ]
  
x _ c(1,2,3)  # Note the use of "_" as a synonym for "<-"
  
y _ c(4,5,6)
  
x+y
  
x*y
  
X _ matrix(scan(),byrow=T,ncol=3)  # note the use of scan()
  
Y _ matrix(sample(1:10,9,replace=T),byrow=T,ncol=3)
  
X
  
Y
  
X %*% Y

dimnames(X) _ list(c("first row","second row","third row"),c("fred",
 "barney","wilma"))

X

dimnames(Y)

dimnames(X)

#############################################################

# [ Numerical Summaries ]

dim(prestige)

for (i in 1:6) print(summary(prestige[,i]))

prestige[1,]

summary(prestige$education)

summary(prestige$income)

summary(prestige$pct.female)

summary(prestige$occ.prestige)

summary(prestige$occ.code)

summary(prestige$occ.type)

sqrt(var(prestige$income))

sqrt(diag(var(prestige[,1:5])))

############################################################

# [ Attach and Detach ]

attach(prestige)

detach()

############################################################

# [ Using Attach and Detach ]

attach(prestige)

stem(education)

stem(income)

stem(pct.female)

stem(occ.code)

stem(occ.type)

dimnames(prestige)[[1]][order(-income)][1:4]

detach()

####################################################################

# [ Graphics, Graphic devices ]

motif()  # or win.graph() or graphsheet() on a PC

par(mfrow=c(2,2))

for (i in c("education","income","pct.female","occ.prestige")) {
 hist(prestige[,i])
 title(i)
 }

postscript("mygraph.ps")

par(mfrow=c(2,2))

for (i in c("education","income","pct.female","occ.prestige")) {
 hist(prestige[,i])
 title(i)
 }

dev.off()

####################################################################

# [ Kernel Density Estimators ]

par(mfrow=c(2,4))

for (i in names(prestige)[1:4]) {
 plot(density(prestige[,i]),type="l")
 title(i)
 }

for (i in names(prestige)[1:4]) {
 plot(density(prestige[,i],width=diff(range(prestige[,i]))/4),type="l")
 title(i)
 }

####################################################################

# [ Boxplots ]

par(mfrow=c(2,4))

for (i in names(prestige)[1:4]) {
 boxplot(prestige[,i])
 title(i)
 }

for (i in names(prestige)[1:4]) {
 boxplot(prestige[,i],style.bxp="old")
 title(i)
 }

####################################################################

# [ boxplots on the same scale ]

par(mfrow=c(1,1))

boxplot(split(prestige$income,prestige$occ.type),style.bxp="old")

attach(prestige)

dimnames(prestige)[[1]][occ.type=="bc"][order(-income[occ.type=="bc"])][1:3]

dimnames(prestige)[[1]][occ.type=="prof"][order(-income[occ.type=="prof"])][1:3]

detach()

###################################################################

# [ Normal probability plot -- several frames of this ] 

par(mfrow=c(3,2))

for (i in 1:6) {
 zz _ qqnorm(rnorm(102))
 abline(lmsreg(zz$x,zz$y))
 }

mtext("Some qqplots of Normal data", outer=T,cex=1.5)

###################################################################

# [ Transformations to achieve normality ]

attach(prestige)

par(mfrow=c(3,2))

plot(density(income),type="l")

points(income,rep(0,102))

zz _ qqnorm(income)

abline(lmsreg(zz$x,zz$y))

plot(density(sqrt(income)),type="l")

points(sqrt(income),rep(0,102))

zz _ qqnorm(sqrt(income))

abline(lmsreg(zz$x,zz$y))

plot(density(log(income)),type="l")

points(log(income),rep(0,102))

zz _ qqnorm(log(income))

abline(lmsreg(zz$x,zz$y))

###################################################################

# [ Box-Cox transformations ]

par(mfrow=c(3,1))

boxplot(split(income,occ.type),style.bxp="old")

title("income")

boxplot(split((income^.25-1)/.25,occ.type),style.bxp="old")

title("(income^.25-1)/.25")

boxplot(split(log(income),occ.type),style.bxp="old")

title("log(income)")

###################################################################

# [ transforming a percent ]

stem(pct.female)

range(pct.female/100)

newpct _ 0.001 + .998*pct.female/100

range(newpct)

stem(log(newpct/(1-newpct)))

stem(qnorm(newpct))

stem(asin(sqrt(newpct)))

###################################################################

[ brush/spin and scatterplot matrices ]