Data Transformations

# =====================================================================================================================
# TRANSFORMATIONS
# =====================================================================================================================

# Focus our attention on a subset of the baseball data.
#
baseball = select(baseball, Name, Atbatc:Walksc)

# Box plots.
#
par(mfrow = c(1, 6), mar = c(4, 3, 3, 1) + 0.1)
for (n in 2:7) {
  boxplot(baseball[,n], xlab = colnames(baseball)[n], col = "lightgreen")
}

# Histograms.
#
par(mfrow = c(1, 6), mar = c(4, 3, 3, 1) + 0.1)
for (n in 2:7) {
  hist(baseball[,n], main = colnames(baseball)[n], col = "lightblue")
}

atbatc.sqrt <- sqrt(baseball[,2])
atbatc.log  <- log10(baseball[,2])
#
boxcox = BoxCoxTrans(baseball[,2])
atbatc.boxcox <- predict(boxcox, baseball[,2])

# Let's focus on one variable.
#
par(mfrow = c(1, 4), mar = c(4, 3, 3, 1) + 0.1)
hist(baseball[,2], col = "lightblue", main = "Atbatc")
hist(atbatc.sqrt, col = "lightblue", main = "sqrt()")
hist(atbatc.log, col = "lightblue", main = "log10()")
hist(atbatc.boxcox, col = "lightblue", main = "Box-Cox")

shapiro.test(baseball[,2])
shapiro.test(atbatc.sqrt)
shapiro.test(atbatc.log)
shapiro.test(atbatc.boxcox)

qqnorm(atbatc.boxcox)
qqline(atbatc.boxcox)

qqnorm(baseball[,2])
qqline(baseball[,2])