# ------------------------------------------------
# Program: Univariate2.R  
#  Author: Steven M. Boker
#    Date: Tue Sep 18 11:11:24 EDT 2007
#
# Provides examples of the use of scatterplots, boxplots and
# histograms using R syntax.
#
# ------------------------------------------------

# ------------------------------------------------
# Create two data vectors and a dataframe of them. 

X <- c(1,2,3,4,5,6,7,8,9,10)

Y <- c(2,4,3,5,4,6,5,7,6,8)

Test2 <- data.frame(X=X, Y=Y)

Test2

summary(Test2)
dim(Test2)


# ------------------------------------------------
# Scatter plot. 

pdf("Univariate2Scatter1.pdf", height=5, width=5)
plot(X, Y)
dev.off()

# ------------------------------------------------
# Scatter plot with fixed axis scales. 

pdf("Univariate2Scatter2.pdf", height=5, width=5)
plot(c(0,10),c(0,10),
     main = "Test 2", 
     ylab = "Y",
     xlab = "X", 
     type="n")
lines(X, Y, type="p")
dev.off()

# ------------------------------------------------
# Read the iris data. 

iris <- data.frame(read.table("iris.dat", header=T))
summary(iris)
dim(iris)


# ------------------------------------------------
# Box plot of Iris sepal length. 

pdf("Univariate2Boxplot1.pdf", height=5, width=6)
boxplot(iris$sepal.length, 
        ylab="sepal.length",
        col="light green")
dev.off()

# ------------------------------------------------
# Multiple boxplots on the same page. 

pdf("Univariate2Boxplot2.pdf", height=7, width=7)
par(mfrow=c(2,2))
boxplot(iris$sepal.length, 
        ylab="sepal.length",
        col="light green")
boxplot(iris$sepal.width, 
        ylab="sepal.width",
        col="light green")
boxplot(iris$petal.length, 
        ylab="petal.length",
        col="light green")
boxplot(iris$petal.width, 
        ylab="petal.width",
        col="light green")
dev.off()

# ------------------------------------------------
# Multiple boxplots in the same graph. 

pdf("Univariate2Boxplot3.pdf", height=5, width=6)
boxplot(list(iris$sepal.length, iris$sepal.width, iris$petal.length, iris$petal.width),
        names=c("sepal.length", "sepal.width", "petal.length", "petal.width"),
        col=c("red", "blue", "green", "purple"))
dev.off()

pdf("Univariate2Boxplot3a.pdf", height=5, width=6)
boxplot(list(iris$sepal.length, iris$sepal.width, iris$petal.length, iris$petal.width),
        names=c("sepal.length", "sepal.width", "petal.length", "petal.width"),
        col=heat.colors(4))
dev.off()

pdf("Univariate2Boxplot3b.pdf", height=5, width=6)
boxplot(list(iris$sepal.length, iris$sepal.width, iris$petal.length, iris$petal.width),
        names=c("sepal.length", "sepal.width", "petal.length", "petal.width"),
        col=c("skyblue1"))
dev.off()

colors()

# ------------------------------------------------
# Splitting on a variable to create multiple boxplots. 

pdf("Univariate2Boxplot4.pdf", height=5, width=6)
boxplot(split(iris$sepal.length, as.factor(iris$variety)), 
        ylab="sepal.length")
dev.off()

# ------------------------------------------------
# Putting it all together. 

pdf("Univariate2Boxplot5.pdf", height=7, width=7)
par(mfrow=c(2,2))
boxplot(split(iris$sepal.length, as.factor(iris$variety)), 
        main="a",
        ylab="sepal.length")
boxplot(split(iris$sepal.width, as.factor(iris$variety)), 
        main="b",
        ylab="sepal.width")
boxplot(split(iris$petal.length, as.factor(iris$variety)), 
        main="c",
        ylab="petal.length")
boxplot(split(iris$petal.width, as.factor(iris$variety)), 
        main="d",
        ylab="petal.width")
dev.off()

# ------------------------------------------------
# ------------------------------------------------
# ------------------------------------------------
# Histogram of sepal length. 

pdf("Univariate2Histogram1-4.pdf", height=5, width=6)
hist(iris$sepal.length, 
     nclass=4, 
     xlab="sepal.length", 
     col="aquamarine", 
     main="")
dev.off()

pdf("Univariate2Histogram1-8.pdf", height=5, width=6)
hist(iris$sepal.length, 
     nclass=8, 
     xlab="sepal.length", 
     col="aquamarine", 
     main="")
dev.off()

pdf("Univariate2Histogram1-16.pdf", height=5, width=6)
hist(iris$sepal.length, 
     nclass=16, 
     xlab="sepal.length", 
     col="aquamarine", 
     main="")
dev.off()

pdf("Univariate2Histogram1-32.pdf", height=5, width=6)
hist(iris$sepal.length, 
     nclass=32, 
     xlab="sepal.length", 
     col="aquamarine", 
     main="")
dev.off()

pdf("Univariate2Histogram1-64.pdf", height=5, width=6)
hist(iris$sepal.length, 
     nclass=64, 
     xlab="sepal.length", 
     col="aquamarine", 
     main="")
dev.off()

# ------------------------------------------------
# Histogram of sepal length with density curve. 

pdf("Univariate2HistogramDensity1.pdf", height=5, width=6)
tdensity <- density(iris$sepal.length)
thist <- hist(iris$sepal.length, 
              nclass=10, 
              xlab="sepal.length", 
              freq=FALSE, 
              main="", 
              col="salmon")
lines(tdensity$x, tdensity$y, type='l')
dev.off()

# ------------------------------------------------
# Multiple histograms with density curves. 

pdf("Univariate2HistogramDensity2.pdf", height=8, width=8)
par(mfrow=c(2,2))
tdensity <- density(iris$sepal.length)
thist <- hist(iris$sepal.length, 
              nclass=10, 
              xlab="sepal.length", 
              freq=FALSE, 
              main="a", 
              col="seagreen1")
lines(tdensity$x, tdensity$y, type='l')
tdensity <- density(iris$sepal.width)
thist <- hist(iris$sepal.width, 
              nclass=10, 
              xlab="sepal.width", 
              freq=FALSE, 
              main="b", 
              col="seagreen1")
lines(tdensity$x, tdensity$y, type='l')
tdensity <- density(iris$petal.length)
thist <- hist(iris$petal.length, 
              nclass=10, 
              xlab="petal.length", 
              freq=FALSE, 
              main="c", 
              col="seagreen1")
lines(tdensity$x, tdensity$y, type='l')
tdensity <- density(iris$petal.width)
thist <- hist(iris$petal.width, 
              nclass=10, 
              xlab="petal.width", 
              freq=FALSE, 
              main="d", 
              col="seagreen1")
lines(tdensity$x, tdensity$y, type='l')
dev.off()