#  Day 2 R scripts
############################################

#############################################
# Salary Data                               #
#############################################

# You should first read the data in... either using the following command
# or use the "Import Dataset" buttom 
#SalaryData = read.table("SalaryData.txt",header=T)

attach(SalaryData)

plot(Gender,Salary,col="red",lwd=1.5,main="Salary Data")

plot(YrHired,Salary,xlab="Year Hired")
points(YrHired[Gender=="Male"],Salary[Gender=="Male"],col="blue",pch=19)
points(YrHired[Gender=="Female"],Salary[Gender=="Female"],col="red",pch=19)

plot(YrsPrior,Salary)
points(YrHired[Gender=="Male"],Salary[Gender=="Male"],col="red",pch=19)
points(YrHired[Gender=="Female"],Salary[Gender=="Female"],col="blue",pch=19)


model1 = lm(Salary~Gender)

model2 = lm(Salary~Gender+YrHired)
plot(YrHired,Salary,xlab="Year Hired")
abline(coef=model2$coef[c(1,3)],col="red",lwd=2)
points(YrHired[Gender=="Female"],Salary[Gender=="Female"],col="red",pch=19)
abline(coef=c(model2$coef[1]+model2$coef[2],model2$coef[3]),col="blue",lwd=2)
points(YrHired[Gender=="Male"],Salary[Gender=="Male"],col="blue",pch=19)


model3 = lm(Salary~Gender*YrHired)
plot(YrHired,Salary,xlab="Year Hired")
abline(coef=model3$coef[c(1,3)],col="red",lwd=2)
points(YrHired[Gender=="Female"],Salary[Gender=="Female"],col="red",pch=19)
abline(coef=c(model3$coef[1]+model3$coef[2],model3$coef[3]+model3$coef[4]),col="blue",lwd=2)
points(YrHired[Gender=="Male"],Salary[Gender=="Male"],col="blue",pch=19)



## Another Housing Data
#######################

#MidCity = read.table("MidCity.txt",header=T)
attach(MidCity)

n = dim(MidCity)[1]

dn1 = rep(0,n)
dn1[Nbhd==1]=1

dn2 = rep(0,n)
dn2[Nbhd==2]=1

dn3 = rep(0,n)
dn3[Nbhd==3]=1

BR = rep(0,n)
BR[Brick=="Yes"]=1


Price = Price/1000
SqFt = SqFt/1000


MidCityModel = lm(Price~dn1+dn2+SqFt)

plot(SqFt,Price,xlab="Size")
abline(coef=MidCityModel$coef[c(1,4)],col=2,lwd=2)
points(SqFt[dn3==1],Price[dn3==1],col=2,pch=19)
abline(coef=c(MidCityModel$coef[1]+MidCityModel$coef[2],MidCityModel$coef[4]),col=4,lwd=2)
points(SqFt[dn1==1],Price[dn1==1],col=4,pch=19)
abline(coef=c(MidCityModel$coef[1]+MidCityModel$coef[3],MidCityModel$coef[4]),col=3,lwd=2)
points(SqFt[dn2==1],Price[dn2==1],col=3,pch=19)
legend(1.45,210,c("Nbhd = 1","Nbhd = 2","Nbhd = 3","Just Size"),lty=c(1,1,1,2),lwd=c(2,2,2,2),col=c(4,3,2,1))
abline(lsfit(SqFt,Price),col=1,lwd=2,lty=2)


Nbhd = factor(Nbhd)

model2 = lm(Price~Nbhd + SqFt + Nbhd:Brick)