#Inputting wage data
df <- read.csv("wage.csv", header = TRUE)

#Giving our variables shorter names
wage <- df$WAGE 
ed <- df$ED 
hsd <- df$HSDROP
hsg <- df$HSGRAD
assc <- df$ASSOC
bach <- df$BACH
age <- df$AGE  
fmale <- df$FEMALE 
nonwh <- df$NONWH
hisp <- df$HISP

#Getting summary statistics for different variables
summary(df)
summary(wage)
summary(df$WAGE)

#Plotting different variables by wage
plot(df$ED,df$WAGE)
plot(ed,wage, col = "blue", pch = 4)
plot(fmale,wage)
plot(age,wage)

#Plotting wage vs age by sex
plot(age,wage,pch=c(2,8)[unclass(df$SEX)],col=c("red","blue")[unclass(df$SEX)])
legend("topleft",legend = c("Female", "Male"), col = c("red","blue"), pch = c(2,8))

#Plotting wage vs age by educational attainment
plot(age,wage,pch=c(2,8,1,3)[unclass(df$EDSTR)],col=c("orange", "purple", "black","blue")[unclass(df$EDSTR)])
legend("topleft",legend = c("hsd", "hsg", "assoc", "bach"), col = c("black", "blue","orange", "purple"), pch = c(1,3,2,8))

#Clearing out our environment
remove(list = ls())
#Clearing all plots
dev.off(dev.list()["RStudioGD"])
#Clearing the console
cat("\014")

#Gather another data set to play with
#Inputting the data from Google Trends, and naming the variables
trends <- read.csv("multiTimeline.csv",header=T)
batman <- trends$Batman
superman <- trends$Superman
wonderwoman <- trends$WonderWoman

#Plotting the variables together on one plot
plot(batman,type='l',col="black")
lines(superman,col="blue")
lines(wonderwoman,col="darkgoldenrod")

#Adjusting the axis and recreating the plot
plot(batman,type='l',col="black", ylim = c(0,100))
lines(superman,col="blue")
lines(wonderwoman,col="darkgoldenrod")

#Plotting the densities of each variable together - need to plot Superman first, since it has a higher mode
plot(density(wonderwoman),col="darkgoldenrod",xlim=c(0,100), main = "Superhero Densities")
lines(density(superman), col="blue")
lines(density(batman))

#Use the par command to put 4 plots together on one page
par(mfrow=c(1,2))
plot(batman,type='l',col="black", ylim = c(0,100), main = "Superhero Scores", ylab = "Trend Score")
lines(superman,col="blue")
lines(wonderwoman,col="darkgoldenrod")

plot(density(wonderwoman),col="darkgoldenrod",xlim=c(0,100), main = "Superhero Densities", xlab = "Trend Score")
lines(density(superman), col="blue")
lines(density(batman))