# Topic 12 How Faithful is Old Faithful?

# R commands for Question 4.

data=read.table("http://bayes.bgsu.edu/m400/data/oldfaith.txt",
  header=T,sep="\t")
attach(data)

# Part (a)
# creates a categorical variable for eruption length

DurLCat=cut(DurLast,c(0,2.5,Inf),c("short","long"))

# finds descriptive statistics for short and long eruptions

summary(DurLast[DurLCat=="short"])
summary(DurLast[DurLCat=="long"])

# Part (b)
# compare the waiting times for short and long eruptions

boxplot(WaitNext~DurLCat,ylab="Waiting Time")

# Part (c)
# creates a categorical variable for waiting time

WaitNCat=cut(WaitNext,c(0,65,Inf),c("short","long"))
summary(WaitNext[WaitNCat=="short"])
summary(WaitNext[WaitNCat=="long"])

# Part (d)
# creates a cross tabulation of two categorical variables

crosstab=table(DurLCat, WaitNCat)

# Part (e)
# are eruption length and waiting time independent?

chisq.test(crosstab)