library(ggplot2)

head(diamonds)
str(diamonds)
?diamonds

summary(diamonds)

# Examining distributions
# ======================================================

# Histograms ---------------------------------------
qplot(price, data=diamonds, geom="histogram", fill=I("black")) 

# ALWAYS EXPERIMENT WITH THE BIN SIZE!
qplot(price, data=diamonds, geom="histogram", binwidth=500, fill=I("black"))
# ALWAYS EXPERIMENT WITH THE BIN SIZE!
qplot(price, data=diamonds, geom="histogram", binwidth=100, fill=I("black"))
# ALWAYS EXPERIMENT WITH THE BIN SIZE!
qplot(price, data=diamonds, geom="histogram", binwidth=50, fill=I("black"))

# Zoom in
qplot(price, data=diamonds, geom="histogram", binwidth=50, fill=I("black"),
      xlim=c(0,2500))

# Your turn 1
qplot(carat, data=diamonds, geom="histogram", fill=I("black")) 
qplot(carat, data=diamonds, geom="histogram", binwidth=0.2, fill=I("black"))
qplot(carat, data=diamonds, geom="histogram", binwidth=0.1, fill=I("black"))
qplot(carat, data=diamonds, geom="histogram", binwidth=0.05, fill=I("black"))
qplot(carat, data=diamonds, geom="histogram", binwidth=0.01, fill=I("black"))

# Investigating relationships
# ======================================================

# Two continuous variables -----------------------------------
# Use a scatterplot
qplot(carat, price, data=diamonds)
qplot(log(carat), log(price), data=diamonds)
qplot(carat, price/carat, data=diamonds)

# Map extra variables to other aesthetic attributes
qplot(carat, price, data=diamonds, colour=clarity)

# Categorical and continuous
qplot(color, price/carat, data=diamonds)
qplot(color, price/carat, data=diamonds, geom="jitter")
qplot(color, price/carat, data=diamonds, geom="boxplot")
qplot(color, price/carat, data=diamonds, geom=c("jitter","boxplot"))
qplot(color, price/carat, data=diamonds, geom="jitter",
  colour=I(alpha("grey60",0.5))) + 
  geom_boxplot(colour=I("black"),fill=I("red"))

# Facetting displays the same plot for different subsets of the data
# use facets argument, rows on left hand-size 
qplot(carat, price, data=diamonds, facets = . ~ color)
qplot(carat, price, data=diamonds, facets = . ~ color,
      colour=I(alpha("black",0.1)))
qplot(carat, price, data=diamonds, facets = . ~ color,
      colour=I(alpha("black",0.1))) + geom_smooth(colour=I("red"), method="lm")

qplot(log(carat), log(price), data=diamonds,
      facets = . ~ clarity,
      colour=I(alpha("black",0.1))) + geom_smooth(colour=I("red"), method="lm")

# Or we can use a histogram for each colour to look at the shape of
# the distribution in more detail:
qplot(price, data=diamonds, facets= color ~ ., geom="histogram")

# ALWAYS EXPERIMENT WITH THE BIN SIZE!
qplot(price/carat, data=diamonds, facets= color ~ ., geom="histogram", binwidth=100)

# Two categorical variables -----------------------------------
# Use a fluctuation diagram - a visualisation of the contingency table
qplot(color, cut, data=diamonds, geom="jitter", colour=I(alpha("black", 0.1)))
ggfluctuation(table(diamonds$cut, diamonds$color))
ggfluctuation(table(diamonds$cut, diamonds$color), type="colour")

# row and column sums:
diamonds$cut <- factor(diamonds$cut, c("Fair","Good","Very Good", "Premium","Ideal"))
diamonds$clarity <- factor(diamonds$clarity, c("IF", "VVS1", "VVS2", "VS1", "VS2", "SI1", "SI2", "I1"))
cut_color <- table(diamonds$cut, diamonds$color)
names(dimnames(cut_color)) <- c("cut", "color")
ggfluctuation(cut_color)
ggfluctuation(cut_color / colSums(cut_color))
ggfluctuation(cut_color / rowSums(cut_color))

# Focus only on best/worst clarity, best/worst color
diamonds.sub<-subset(diamonds, clarity=="I1" | clarity=="IF")
diamonds.sub<-subset(diamonds.sub, color=="J" | color=="D")

dim(diamonds.sub)
qplot(log(carat), log(price), data=diamonds.sub,
      facets = color ~ clarity,
      colour=I("black")) + geom_smooth(colour=I("red"), method="lm")
qplot(log(carat), log(price), data=diamonds.sub,
      facets = cut ~ clarity+color,
      colour=I("black")) + geom_smooth(colour=I("red"), method="lm")

# Draw multiple lines on the same scatterplot
p <- qplot(log(carat), log(price), data=diamonds, geom="point")
p + geom_smooth(aes(x=log(carat), y=log(price), colour=color), method="lm")
p <- qplot(log(carat), log(price), data=diamonds, geom="point")
p + geom_smooth(aes(x=log(carat), y=log(price), colour=clarity), method="lm")