t4q1 <- read.table("C:/Users/Wei Hao/Desktop/ST2137/Tutorials/Data/furniture.txt",header=T)
attach(t4q1)
IQR(days)
## [1] 38.25
mad(days, constant=1)
## [1] 15.5
mad(days)
## [1] 22.9803
The three robust estimators are smaller than the s.d. It seems that there are some big values in the data set that inflated the s.d. The Gini’s mean difference try to reduce the effect of the extreme values but the result is not as good as the other two robust estimators of \(\sigma\).
The variation is better measured by the MAD. The estimate for \(\sigma\) based on the MAD is about \(22\). All these estimators are robust against outliers in the data set.
t4q2 <- read.table("C:/Users/Wei Hao/Desktop/ST2137/Tutorials/Data/student.txt",header=T)
attach(t4q2)
gendergp <- ifelse(gender=="F","Female","Male")
table(gendergp)
## gendergp
## Female Male
## 96 123
travelgp <- ifelse(travel=="Y","Yes","No")
table(travelgp)
## travelgp
## No Yes
## 53 166
drivelicgp <- ifelse(drivelic=="Y","Yes","No")
table(drivelicgp)
## drivelicgp
## No Yes
## 78 141
table(gendergp, drivelicgp)
## drivelicgp
## gendergp No Yes
## Female 36 60
## Male 42 81
chisq.test(table(gendergp, drivelicgp))
##
## Pearson's Chi-squared test with Yates' continuity correction
##
## data: table(gendergp, drivelicgp)
## X-squared = 0.13842, df = 1, p-value = 0.7099
Since p-value \(=0.7099 > 0.05\), we do not reject the null and conclude that they are independent.
wkhrgp = character(length(workhour))
wkhrgp[workhour == 0] = "None (0 hrs)"
wkhrgp[workhour > 0 & workhour < 20] = "Some (1 - 19 hrs)"
wkhrgp[workhour >= 20] = "Many (20 - 99 hrs)"
table(wkhrgp)
## wkhrgp
## Many (20 - 99 hrs) None (0 hrs) Some (1 - 19 hrs)
## 37 101 81
chisq.test(table(wkhrgp, travelgp))
##
## Pearson's Chi-squared test
##
## data: table(wkhrgp, travelgp)
## X-squared = 5.8444, df = 2, p-value = 0.05382
Since the observed \(\chi^2\) value is \(5.8444 < \chi_{0.05}^2 (2) = 5.9915\) (or with p-value = 0.0538), we do not reject \(H_0\) and conclude that travel and wkhrgp are independent.
v <- matrix(c(16, 24, 654, 306), nc = 2,byrow=T)
chisq.test(v, correct=T)
##
## Pearson's Chi-squared test with Yates' continuity correction
##
## data: v
## X-squared = 12.496, df = 1, p-value = 0.0004079
chisq.test(v, correct=FALSE)
##
## Pearson's Chi-squared test
##
## data: v
## X-squared = 13.738, df = 1, p-value = 0.0002101
Since the observed $^2 $ value \(=13.74 > \chi_{0.05}^2 = 3.8415\) (or with p-value \(= 0.0002\)), we reject \(H_0\) that “conform” and “shift” are independent.