#Cholesterol Data Analysis:

#Read the data:
cholesterol=read.table("cholesterol.txt",header=T)

#Extract the variables:
sex=cholesterol[,"SEX"]
age=cholesterol[,"AGE"]
tc=cholesterol[,"TC"]
hdl=cholesterol[,"HDL"]
trig=cholesterol[,"TRIGLYCERIDE"]
ldl=cholesterol[,"LDL"]

#Scatter plot of the data:
plot(cholesterol)
par(ask=T)

par(mfrow=c(2,4))
#Boxplots:
boxplot(tc~sex, main="TC")
boxplot(hdl~sex, main="HDL")
boxplot(trig~sex, main="TRIG")
boxplot(ldl~sex, main="LDL")

boxplot(tc~age)
boxplot(hdl~age)
boxplot(trig~age)
boxplot(ldl~age)

#Histograms:
hist(tc)
hist(hdl)
hist(trig)
hist(ldl)

hist(log(tc))
hist(log(hdl))
hist(log(trig))
hist(log(ldl))

#Define new measures:
new=100*log(hdl)/(log(hdl)+log(ldl))

#Plot new variable:
par(mfrow=c(2,2))
hist(new)
plot(hdl+ldl,new,main="new vs. old")
boxplot(new~sex, main="NEW")
boxplot(new~age, main="NEW")

plot(hdl,new)
plot(ldl,new)
plot(tc,new)
plot(trig,new)

sink("CholesterolAnalysis.txt")
#Get summary statistics:
print(summary(cholesterol[,-1]))
print(summary(new))
print(summary(lm(new ~ ldl+hdl)))
print(summary(lm(new ~ tc)))
print(summary(lm(new ~ trig)))
sink()