Multivariate data

In this chapter, we are going to describe relationships, and begin working with multivariate data, which is a fancy way of saying samples containing more than one variable.

The troublemaker reader might remark that all the datasets that we've worked with thus far (mtcars and airquality) have contained more than one variable. This is technically true—but only technically. The fact of the matter is that we've only been working with one of the dataset's variables at any one time. Note that multivariate analytics is not the same as doing univariate analytics on more than one variable–multivariate analyses and describing relationships involve several variables at the same time.

To put this more concretely, in the last chapter we described the shape of, say, the temperature readings in the airquality dataset.

  > head(airquality)
  Ozone Solar.R Wind Temp Month Day
  1    41     190  7.4   67     5   1
  2    36     118  8.0   72     5   2
  3    12     149 12.6   74     5   3
  4    18     313 11.5   62     5   4
  5    NA      NA 14.3   56     5   5
  6    28      NA 14.9   66     5   6

In this chapter, we will be exploring whether there is a relationship between temperature and the month in which the temperature was taken (spoiler alert: there is!).

The kind of multivariate analysis you perform is heavily influenced by the type of data that you are working with. There are three broad classes of bivariate (or two variable) relationships:

• The relationship between one categorical variable and one continuous variable
• The relationship between two categorical variables
• The relationship between two continuous variables

We will get into all of these in the next three sections. In the section after that, we will touch on describing the relationships between more than two variables. Finally, following in the tradition of the previous chapter, we will end with a section on how to create your own plots to capture the relationships that we'll be exploring.