In ggplot2, aesthetic means “something you can see”. Every layer has a specific set of aesthetic variables. These parameters provide all information to draw the graph and all its components. The data (vectors) passed to the aesthetic parameters must have the same length as the amount of rows in the given dataframe or must be of length 1. Passing a vector of length 1 will define the aesthetic for all rows in the dataframe.

First, we need to understand that any aesthetic in ggplot2 (such as x, y, colour, size, shape, etc.) can be defined in two distinct ways in your plots: setting and mapping.

setting aesthetics

In most cases you will set aesthetics when you want to set a constant aesthetic value for ALL data entries (vector of length one)

Setting an aesthetic can be done by passing them as a parameter to the ggplot (layer) functions.

murders %>% ggplot() +
    geom_point(x=murders$population/10^6, y=murders$total)

• The disadvantage is that your in full controll of the graph, you need to specify every graphical feature by yourself. Let’s set the the geom_point x and y aesthetics, an empty plot appears:

  murders %>% ggplot() +
    geom_point(x=murders$population/10^6, y=murders$total)
  

  

  This is because there are no scales defined, after defining the scales and axis labels we see the expected plot appear:

  murders %>% ggplot() +
    geom_point(x=murders$population/10^6, y=murders$total) +
    # add axis scales and labels
    scale_y_continuous(limits= c(0,max(murders$total)), name = "total") +
    scale_x_continuous(limits= c(0,max(murders$population/10^6)), name = "population/10^6")
  

  

• An advantage of setting aesthetics is the simplycity for setting a constant aesthetic for all datapoints. To make all the datapoints blue we can set the colour aesthectic to "blue":

  murders %>% ggplot() +
    geom_point(x=murders$population/10^6, y=murders$total, colour="blue") +
    # add axis scales and labels
    scale_y_continuous(limits= c(0,max(murders$total)), name = "total") +
    scale_x_continuous(limits= c(0,max(murders$population/10^6)), name = "population/10^6")
  

  

mapping aesthetics

In most cases you will map aesthetics when you want to map every data entry with it’s own aesthetic value.

To map aesthetics you will need to define the aesthetics in the aes() function and not directly in the ggplot (layer) function. The aes() function will then return the aesthetic mapping which can be passed to the ggplot (layer) functions.

aes(x = murders$population/10^6, y = murders$total)
#> Aesthetic mapping:
#> * `x` -> `murders$population`
#> * `y` -> `murders$total`

The aes function inside a ggplot (layer) function also uses the variable names from the given dataframe: we can use population and total without having to call them as murders$population and murders$total (Same as dplyr functions: filter, mutate, select, etc..).

murders %>% ggplot() +
  geom_point(aes(x = population/10^6, y = total))

We can drop the x= and y= since these are the first and second expected arguments, you can find this info in the help page.

murders %>% ggplot() +
  geom_point(aes(population/10^6, total))

• An advantage of using the aesthetic mapping is that behind the screens it does much more then setting the aesthetic. The scale and labels are defined by automatically when adding this layer:

  murders %>% ggplot() +
    geom_point(aes(population/10^6, total))
  

  

• A drawback of the aesthetic mapping is that it sometimes does more then you want for simple tasks. When we map the colour aethetic to "blue" the geom_point layer interprets this as a categorical variable (not as the colour blue) and gives all points a color depending on its category. The geom_point layer will choose the colour automatically, in this case the default is red. If there were more categories then ggplot would automatically assign a different colour to each category.

  murders %>% ggplot() +
    geom_point(aes(population/10^6, total, colour="blue"))
  

  

  To fix this, we can instead set the colour aesthetic to “blue” by specifying it outside the aes function.

  murders %>% ggplot() +
    geom_point(aes(population/10^6, total), colour="blue")
  

  

Conclusion

Despite being equivalent, both setting and mapping have advantages and disadvantages. The key is to combine these methods depending on your goals.