Exercise

Graphical tests for multivariate normality

You are often required to verify that multivariate data follow a multivariate normal distribution. Recall that univariate normality of each individual variable does not imply multivariate normality overall. We will use functions to check multivariate normality of all variables instead of univariate normality of single variables.

Note, that many of functions discussed in these exercises are deprecated in the most recent MVN library. If you wish to use these functions install the archived version using

install.packages("devtools")
library(devtools)
install_version('MVN', '4.0.2')

.

Instructions 1/3

undefined XP
  • 1

    Use the qqplot() and qqline() functions to check the univariate normality of the first column of multnorm.sample.

  • 2

    Now, use the uniPlot() function to draw the univariate qqplots of the first four numeric variables, Alcohol, Malic, Ash, Alcalinity of the wine dataset.

  • 3

    Use the mardiaTest() function to draw the QQ-plot to test for multivariate normality for the first four numeric variables of the wine dataset.