• Brushing: highlight categories by selecting them
• Linking: link selections across several plots
• Animation: selections in sequence; ideally can be played automatically or controlled by user

Good design still aids comprehension and usability

• You need your story to be clear and readable

• Details can be hidden
• Structure and logic should be visible

Why don’t people do this more?

In what contexts can you reach an audience with interactive graphics?

How can interactive graphics help you with your research?

Interactive graphics can be very useful for certain kinds of exploration

Principles of friendliness may no longer apply, since users will become very familiar with the tool

• It’s possible to mislead yourself

You may want to lead your readers through an interactive visualization

Several versions of your plot

Or even several default views of the same plot

What do people need to compare?

• Choose scales
• Make information persistent or retrievable

Schoolchildren example

Tax example

Linking example with cars

gapminder animation

Except for exploration, you don’t want graphs that need to be looked over piece by piece

But it can be a useful option

Many packages just provide this

Someone using interactively does not need to see more than 3 or 4 significant figures

They also do not need to see everything in your data frame

You should not drop precision before making the plot, though

Solution: create dummy columns and limit what is shown to users on labels

Pick out a particular data series

Needs good communication and usability

Particularly useful if combined with linking

Best when it’s under user control

A really good brushing interface can often be better

Can be really nice for talks, though

• Straightforward and popular (esp. via ggplotly)

• very “gg” (grammar of graphics)-ish; not so interactive

• for letting different html widgets talk to each other (linking)

• maps (more later)