#Load and modify data
library(ggplot2)
library(dplyr) 

View(diamonds)
  #only use subset of color and clarity categories 
diamonds2 <-diamonds[diamonds$color %in% c("D","E","F"),]
diamonds2 <-diamonds2[diamonds2$clarity %in% c("SI1","SI2","VS1","VS2"),]
diamonds2 <-droplevels(diamonds2)
  #use log price rather than raw price
diamonds2$ln_price <-round(log(diamonds2$price),2)
View(diamonds2)
#-------------------------------------------------------------------------------


#Determine Variable types and get overall shape of data
summary(diamonds2)

  #Y-variables
summary(diamonds2$ln_price)

  #X-variables
    #continuous variables
summary(diamonds2$carat)
summary(diamonds2$depth)

    #discrete variables
table(diamonds2$cut)
table(diamonds2$color)
table(diamonds2$clarity)

    #discrete variables by price
tapply(diamonds2$ln_price, diamonds2$cut, mean)
tapply(diamonds2$ln_price, diamonds2$color, mean)
tapply(diamonds2$ln_price, diamonds2$clarity, mean)

    #discrete variables by each other
table(diamonds2$cut, diamonds2$color)
table(diamonds2$cut, diamonds2$clarity)
#-------------------------------------------------------------------------------


#Determine formula and get Dirty plots
plot(diamonds2$ln_price ~ diamonds2$carat)
plot(diamonds2$ln_price ~ diamonds2$carat, col=diamonds2$color)

boxplot(diamonds2$ln_price ~ diamonds2$cut)
boxplot(diamonds2$ln_price ~ diamonds2$color + diamonds2$clarity)
#-------------------------------------------------------------------------------


#Get Clean plots and Publication considerations
p1 <-ggplot(data=diamonds2, aes(y=ln_price, x=carat))+
  geom_point() + geom_smooth(method='loess') +
  labs(x="Carot", y="Log price in dollars") 
p1

p2 <-ggplot(data=diamonds2, aes(y=ln_price, x=carat, fill=color, color=color))+
  geom_point() + geom_smooth(method='loess') +
  labs(x="Carot", y="Log price in dollars") 
p2
#-------------------------------------------------------------------------------


d_sum<-diamonds2 %>% 
  group_by(cut) %>%       
  summarise(mean=mean(ln_price),
            sd = sd(ln_price),
            error = qt(0.975,df=n()-1)*sd/sqrt(n()),
            ul = mean + error,
            ll = mean - error)

p3 <-ggplot(data=d_sum, aes(y=mean, x=cut, fill=cut))+
  geom_bar(stat="identity") +
  geom_errorbar(aes(ymin=ll, ymax=ul), width=0.1)+
  labs(y="Mean log price in dollars") 
p3
#-------------------------------------------------------------------------------


d_sum2<-diamonds2 %>% 
  group_by(clarity,color) %>%       
  summarise(mean=mean(ln_price),
            sd = sd(ln_price),
            error = qt(0.975,df=n()-1)*sd/sqrt(n()),
            ul = mean + error,
            ll = mean - error)
p4 <-ggplot(data=d_sum2, aes(y=mean, x=clarity, fill=color)) +
  geom_bar(aes(fill=color), stat="identity", position=position_dodge()) +
  geom_errorbar(aes(ymin=ll, ymax=ul, group=color), width=0.1, 
                position=position_dodge(0.9))+
  labs(y="Mean log price in dollars") 
p4