Effective communication ↔ effective visuals
Difference between a clear message about your data versus a confusing one
Important decisions can hinge on how persuaded people are with your presented work!
Going against plotting conventions, even if the data is literally accurate, can be misleading
Caveat emptor: breaking visual conventions can be a deliberate strategy, approach with caution and careful skepticism
Plot each car’s highway fuel efficiency (
hwy) as a function of the engine size (
ggplot(data = mpg) + geom_point(mapping = aes(x = displ, y = hwy))
ggplot(data = mpg) + geom_point(mapping = aes(x = displ, y = hwy, color = class, shape = factor(cyl)))
We can break visualizations down into four basic elements:
These are the building blocks of any given visualization.
Identify 9 separate visual cues.
Position (numerical) where in relation to other things?
Length (numerical) how big (in one dimension)?
Angle (numerical) how wide? parallel to something else?
Direction (numerical) at what slope? In a time series, going up or down?
Shape (categorical) belonging to which group?
Area (numerical) how big (in two dimensions)?
Volume (numerical) how big (in three dimensions)?
Shade (either) to what extent? how severly?
Color (either) to what extent? how severly? Beware of red/green color blindness.
Cartesian This is the familiar (x, y)-rectangular coordinate system with two perpendicular axes
Polar: The radial analog of the Cartesian system with points identified by their radius ρ and angle θ
Geographic: Locations on the curved surface of the Earth, but represented in a flat two-dimensional plane
Numeric: A numeric quantity is most commonly set on a linear, logarithmic, or percentage scale.
Categorical: A categorical variable may have no ordering or it may be ordinal (position in a series).
Time: A numeric quantity with special properties. Because of the calendar, it can be specified using a series of units (year, month, day). It can also be considered cyclically (years reset back to January, a spring oscillating around a central position).
Annotations and labels that draw attention to specific parts of a visualization.
Axes labels that depict scale (tick mark labels) and indiciate the variable
Reference points or lines
Other markups such as arrows, textboxes, and so on (it’s possible to overdo these)
How many of the previous elements can you identify in this plot?
Split your plot into “facets”; particularly useful for categorical variables.
ggplot(data = mpg) + geom_point(mapping = aes(x = displ, y = hwy)) + facet_wrap( ~ class)
facet_wrap( ~ class, nrow = 2) facet_grid(drv ~ cyl) facet_grid(drv ~ .) facet_grid(. ~ cyl)
Don’t forget the
geom_smooth to dip our toe into the world of data-driven modeling.
What do you get when you run the following?
ggplot(data = mpg) + geom_point(mapping = aes(x = displ, y = hwy)) + geom_smooth(mapping = aes(x = displ, y = hwy))
To use the more familiar linear model (the so-called “line of best fit”), include the input
method = "lm".
ggplot(data = mpg) + geom_point(mapping = aes(x = displ, y = hwy)) + geom_smooth(mapping = aes(x = displ, y = hwy), method = "lm")